《環太平洋大學聯盟：2025高等教育中的生成式人工智能白皮書：現行做法與未來道路（英文版）（40頁）.pdf》由會員分享，可在線閱讀，更多相關《環太平洋大學聯盟：2025高等教育中的生成式人工智能白皮書：現行做法與未來道路（英文版）（40頁）.pdf（40頁珍藏版）》請在三個皮匠報告上搜索。

1、 Generative AI in higher education Current practices and ways forward Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 2 Generative AI in higher education:Current practices and ways forward A whitepaper from the Generative AI in Education:O

2、pportunities,Challenges and Future Directions in Asia and the Pacific project January 2025 Danny Y.T.Liu Professor of Educational Technologies The University of Sydney,Australia Simon Bates Vice-Provost and Associate Vice President,Teaching and Learning The University of British Columbia,Canada Asso

3、ciation of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 3 Contents Contents.3 Foreword.4 Executive summary.6 Introduction.8 Motivation.8 Where is the sector now,and where is it heading?.9 The urgency to act.10 How to use this whitepaper to inform acti

4、on.11 Acknowledgements.12 Five areas for action.14 Immediate key areas of activity.14 Rules.15 Access.19 Familiarity.22 Trust.28 Culture.32 The importance of all five areas.37 Looking ahead.38 Form collaborative clusters.38 Elevate students as partners.39 Conclusion.40 Association of Pacific Rim Uni

5、versities Generative AI in higher education:Current practices and ways forward 4 Foreword This report stakes out a territory in transformation:higher education in the face of the rapid advancement of generative artificial intelligence and its incremental application across all sectors of post-second

6、ary education.Universities as the custodians of this territory are still slow in responding to this change when they should be swift and anticipatory,given the pace of AI development.However,navigating the new reality is complex and requires institutions to rethink basic assumptions that have underp

7、inned their value proposition as education providers and their institutional operations.Universities are faced with an emerging technology that displays still many uncertainties in terms of development,standardization,regulation,and usability.The main takeaways of Stanford Universitys Artificial Int

8、elligence Index Report 20241 demonstrate this clearly:AI has surpassed human performance in some areas but still lags on many more complex tasks.Industry dominates frontier AI research,outdistancing academia and industry-academia collaborations;here,the US outpaces China,the EU,and the UK as the lea

9、ding source of top AI models.The training costs of frontier AI models are increasingly high,while funding for generative AI has surged to reach$25.2 billion annually.Comparing the risks and limitations of top AI models is difficult due to a lack of standardization regarding responsible AI benchmarks

10、.At the same time,AI regulation has seen a significant increase.AI may enhance work productivity and accelerates scientific discovery a spectacular example is Demis Hassabis and John Jumpers breakthrough AI model AlphaFold which allows to predict the structure of virtually all 200 million proteins t

11、hat researchers have identified,recognized by the 2024 Noble Prize in Chemistry2.At the same time,an increasing number of the world population is cognizant of the rising impact of AI on their lives and concerned about it.Many studies contemplate societal benefits and risks to society3.Universities h

12、ave not yet found common ground in how to balance opportunities and risks in the adoption of AI.The 2024 Educause AI Landscape Study4 sees some consensus regarding appropriate uses5 versus inappropriate uses6.Opportunities are mostly seen in improving teaching,learning and student success;data analy

13、tics and access;and relieving administrative workload.Risks associated with the use of AI are mostly located in the areas of ethics(e.g.plagiarism,intellectual property,widening the digital divide,mis-and disinformation),privacy and security,lack of AI literacy,and the threat AI can pose to creativi

14、ty,critical thinking and human engagement in learning.1 Stanford University(2024)The AI index report.https:/aiindex.stanford.edu/report/2 https:/www.nobelprize.org/uploads/2024/10/advanced-chemistryprize2024.pdf 3 European Parliament(2020)https:/www.europarl.europa.eu/topics/en/article/20200918STO87

15、404/artificial-intelligence-threats-and-opportunities;Marr(2023)https:/ Robert(2024)2024 EDUCAUSE AI Landscape Study,https:/library.educause.edu/resources/2024/2/2024-educause-ai-landscape-study.5 Such as personalized student support;use of AI tool as teaching,research and administrative assistant;l

16、earning analytics;digital literacy education.6 Such as trusting generative AI outputs or making high-stakes decisions(e.g.student admissions)without human oversight;simulating human judgment(grading,peer review,writing recommendation letters),representing AI-generated work as ones own,not citing AI

17、as a resource for generated content,conducting invasive data collection or surveillance,relying on AI tools in place of human thought and creativity,giving tools unauthorized access to sensitive data or intellectual property.Association of Pacific Rim Universities Generative AI in higher education:C

18、urrent practices and ways forward 5 On an institutional level,the adoption of AI confronts universities with a range of questions that are fundamental to their identity:Does the educational role and the value proposition of degrees change with the adoption of AI?Do universities maintain authority ov

19、er the education they deliver?How can universities make sure that there is fair and equal access to AI across faculties,programs and curricula?How can the institutional complexity of universities and their inertia be mitigated in an era of increasingly rapid technological change?Will we even witness

20、 a transformation towards news types of universities?The present whitepaper attempts to chart this very territory.It is one of the main outcomes of the APRU project“Generative AI in Higher Education”,conducted by the Association of Pacific Rim Universities(APRU)with the generous support of Microsoft

21、.Following a survey of case studies demonstrating the current use of AI in APRU member universities,three workshops facilitated by Tandemic throughout 2024 including an in-person workshop hosted by The Hong Kong University of Science and Technology in June 2024 brought AI experts together to assess

22、the case studies and to develop scenarios and paradigms of what AI-enhanced universities might look like in 2035.We hope that this whitepaper will make an important contribution to the ongoing debate about the future place of AI in our universities,its promise,and potential.We trust the whitepaper w

23、ill influence policies and support decision-making,thereby promoting a broader reimagination of universities as we enter the second quarter of the 21st century.Let me conclude by extending our warmest gratitude to Microsoft for their most generous sponsorship that has made this project possible.Our

24、special thanks go to Larry Nelson(Asia Regional Business Lead,Education,and General Manager),Madeline Shepherd(Asia Digital Safety Lead)and Lee Hickin(AI Technology and Policy Lead Asia).I also acknowledge Danny Liu and Simon Bates for their expertise and project support as the authors of the whitep

25、aper,as well as Simon Bates additionally for his oversight of the project as the academic project lead.I thank my colleagues Christina Schnleber and Benjamin Zhou from APRU for leading the project development and its implementation,and Kal Joffres from Tandemic for the development and facilitation o

26、f the project workshops.Thomas Schneider,APRU Chief Executive Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 6 Executive summary The wide availability of generative AI represents a pivotal moment for higher education that goes far beyond

27、merely accommodating another technological innovation.It fundamentally challenges our assumptions about teaching,learning,research,and the very purpose of universities.This whitepaper,emerging from collaboration across Pacific Rim universities,presents both a framework for action and a call for tran

28、sformative change in how we prepare students,ourselves,and our institutions for an AI-enabled future.Universities currently face unprecedented pressure to respond to generative AI while maintaining the integrity and value of higher education.Current approaches are typically piecemeal and reactive,fo

29、cusing on immediate concerns like academic integrity rather than systematic integration of AI into educational practice in responsible and productive ways.Meanwhile,students already questioning the value of traditional higher education are embracing AI tools regardless of institutional readiness.Our

30、 sector must move swiftly from policing to possibilities,from panic to purpose.Our work has identified five interdependent elements essential for successful generative AI integration,forming the CRAFT framework culture,rules,access,familiarity,and trust.Culture represents both the deepest challenge

31、and greatest opportunity.Beyond regional and institutional differences in generative AI acceptance and adoption,we must address fundamental questions about the universitys role in an AI-enabled world.Rules must move beyond restriction to enablement,with effective governance frameworks providing clea

32、r guidelines while encouraging experimentation and innovation.Assessment practices particularly require fundamental redesign to ensure both validity and relevance in an AI-enabled world.Access remains a critical equity issue without deliberate intervention,AI risks widening existing digital divides.

33、Institutions must ensure equitable access not just to tools but to the infrastructure,support,and opportunities needed to leverage AI effectively.Familiarity requires systematic development across all stakeholders.Beyond basic digital literacy,we need deep understanding of AI capabilities,limitation

34、s,and ethical implications,demanding sustained investment in development and student support.Trust underpins all progress whether between students and educators,institutions and vendors,universities and their communities,or other trust pairs trust must be actively built and maintained through transp

35、arency,collaboration,and demonstrated value.Individual institutional responses are insufficient for the scale of change required.We propose two key priorities for immediate sector-wide action.First,the formation of collaborative clusters where universities move beyond competition to cooperation in k

36、ey areas including joint development of generative AI applications and pedagogical approaches,shared frameworks for assessment redesign,coordinated advocacy for equitable access,combined faculty development initiatives,and unified governance frameworks that respect local contexts.Second,the elevatio

37、n of students as partners Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 7 through peer-to-peer support networks,student AI ambassador programs,co-design of learning experiences,direct input into assessment redesign,and collaborative reso

38、urce development.The emergence of generative AI may be our best opportunity to reimagine higher education for the 21st century.Success requires us to move beyond incremental adaptation to fundamental transformation while preserving our core educational values.This whitepaper provides a suggested roa

39、dmap,but implementation demands immediate,coordinated action across the sector.We must develop comprehensive institutional AI strategies that address culture,rules,access,familiarity,and trust,working together to address shared challenges and leverage shared opportunities.The choice we face is not w

40、hether to engage with AI but how to shape its integration to enhance rather than diminish the value and transformative power of higher education.The framework and recommendations in this whitepaper provide a foundation for action.The time to act is now.Association of Pacific Rim Universities Generat

41、ive AI in higher education:Current practices and ways forward 8 Introduction Motivation Since ChatGPT was released in November 2022,the higher education sector,industry,and wider society have reacted in very different ways to the implications of generative AI for the present and future.For higher ed

42、ucation an initial moral panic was fuelled by immediate concerns around academic integrity.Since then,there has been a gradual and growing acceptance that generative AI is here to stay and we must adapt to its presence and ever-growing capabilities and the opportunities they present,whilst at the sa

43、me time being cognizant of the challenges and limitations.After all,one of the key groups that higher education serves are its students,whom need to prepare for a world where AI is ubiquitous.However,adaptation and adoption in the higher education sector has generally not been systematic.Artificial

44、intelligence could be considered the newest general purpose technology,an advancement like the steam engine or electricity with impacts across society and the economy.Even though the underlying infrastructure needed(connectivity,software,and hardware)are already largely in place to accelerate adopti

45、on,as with other general-purpose technologies it will take time before its full impact is felt,often because workers and organizations need to learn the technology and adapt organizational processes and structures7.Unlike past general-purpose technologies,the capabilities are advancing rapidly makin

46、g it more challenging to adapt to a fast-moving target.Sector challenges These have certainly compounded the lack of systematic engagement in higher education institutions with generative AI across their education,research,and operational functions.Many institutions lack personnel with necessary exp

47、ertise to implement and manage AI effectively8.There are legitimate concerns around data protection,use and misuse of intellectual property,algorithmic bias,academic integrity,and the ethical and responsible use of AI by students and educators9.Regional differences in regulatory environments contrib

48、ute to uneven access to AI tooling and applications10.Inequitable access and the risk of broadening the digital divide are important considerations,particularly in low-and middle-income countries11.Additionally,an existential threat is felt by higher education researchers and educators who may see t

49、heir functions or parts of their roles being diminished or replaced by AI,may not know how to adapt from more traditional approaches,and are 7 Crafts(2021)Artificial intelligence as a general-purpose technology:an historical perspective.https:/doi.org/10.1093/oxrep/grab012 8 Microsoft(2024)AI in Edu

50、cation:A Microsoft Special Report.http:/aka.ms/AIinEDUReport 9 UNESCO(2023)Guidance for generative AI in education and research.https:/unesdoc.unesco.org/ark:/48223/pf0000386693 10 For example,OpenAI Supported countries and territories:https:/ 11 United Nations(2024)Mind the AI Divide.https:/www.un.

51、org/techenvoy/sites/www.un.org.techenvoy/files/MindtheAIDivide.pdf Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 9 already under significant workload pressures12.Early student perspectives suggest,however,that despite students being open

52、 to receiving assistance from AI,they still value the human elements of teacher-student relationships13.These challenges have led to the cautious and somewhat piecemeal approach to generative AI adoption by universities across institutions comprising the Association of Pacific Rim Universities(APRU)

53、.Like industry,where individual experimentation as opposed to strategic organizational engagement has been the prevailing response14,higher education is now at a stage where it needs to transition to a holistic,supported,and scaffolded approach to generative AI adoption.The higher education sector h

54、as been quick to bring groups together to define and adopt high level principles that espouse humanity,ethics,integrity,amongst others15,but a gulf exists between this and what university stakeholders like leaders,educators,and students need to effectively integrate generative AI into specific educa

55、tional,research,and operational processes.Where is the sector now,and where is it heading?As a component of the project that this whitepaper was developed for,APRU first collated case studies on generative AI use across their member institutions.Supported by the social innovation agency Tandemic,APR

56、U arranged a series of workshops throughout 2024,with input from APRU members and representatives from technology and publishing companies.These workshops sought to discover and share current practice and look to the future of higher education with generative AI in mind.Sensemaking A sensemaking wor

57、kshop(March 2024)identified patterns and trends through case studies of AI use in universities,recognizing gaps and opportunities.16The main insights gained included:(i)the importance of transparency,trust,and culture in AI adoption;(ii)the need to adapt rapidly;(iii)ensuring equitable access to gen

58、erative AI;(iv)how pedagogy needs to drive technological adoption;(v)that universities need to prepare learners for an AI-driven world and shift from a focus on knowledge to values and skills;and(vi)the centrality of human interaction and relationship in higher education.12 Lee et al.(2024)The impac

59、t of generative AI on higher education learning and teaching:A study of educators perspectives.https:/doi.org/10.1016/j.caeai.2024.100221 13 Chan&Tsi(2024)Will generative AI replace teachers in higher education?A study of teacher and student perceptions.https:/doi.org/10.1016/j.stueduc.2024.101395 1

60、4 Relyea et al.(2024)Gen AIs next inflection point.https:/ 15 Australian Government(2024)Study Buddy or Influencer.https:/www.aph.gov.au/Parliamentary_Business/Committees/House/Employment_Education_and_Training/AIineducation/Report 16 APRU(2024)The Future of Generative AI in Higher Education.https:/

61、www.apru.org/our-work/university-leadership/generative-ai-in-education/Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 10 Foresight A foresight workshop(June 2024)explored emerging trends and considered their impact on higher education,cul

62、minating in the creation of models to imagine the future of universities.This workshop highlighted the unprecedented rate and range of disruptions facing the sector,including shifts in perceptions around the value of higher education and employer sentiment.Four models were proposed as provocations f

63、or the future:(i)research collaboratories where students learn through an apprenticeship model and institutions tackle grand global challenges;(ii)the digital university consortia where students learn through a network of experiences from multiple institutions,providing them with marketable skills;(

64、iii)community learning universities which focus on community development and social impact through a small-scale,human-first approach with a diminished role for AI;and(iv)entrenched universities which only change from existing models incrementally and respond slowly to societal and employer expectat

65、ions.Prototypes A creative sandbox workshop(August 2024)turned the models into tangible and testable forms,with the aim of identifying potential issues and opportunities.Five prototypes were developed to test different university models developed in the second workshop,such as the OneUni Alliance wh

66、ere multiple institutions would collaborate to create a personalized learning experience for students who would take multiple courses spanning across different universities.These prototypes allowed the examination of existing policies and practices catalyzed by the disruptive force of generative AI.

67、These included the agility of curriculum processes,encouraging student agency,enabling interdisciplinary learning,integrating learning on AI ethics,evolving roles of faculty,and rethinking institutional governance.The urgency to act The workshops provided a valuable opportunity to share current prac

68、tice and imagine potential futures with a 10+year horizon,uncovering key considerations that institutions need to grapple with right now to prepare for the future.This whitepaper connects some of the shared practices,imagined futures,and emerging considerations with the current Pacific Rim context a

69、nd short-to medium-term actions that institutions should be taking.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 11 One key immediate challenge facing higher education is the integrity of awarded qualifications.With generative AI increas

70、ingly able to perform well in assessments17,unsupervised assessments are no longer able to assure attainment of learning outcomes.This does not mean that every assessment must now be supervised;rather,it means that assessment redesign is needed so that there is a pedagogically beneficial mixture of

71、secured assessment of learning and open assessment for learning.A more medium-term challenge is the need to adequately prepare students for the workforce.Organizations are adopting generative AI at an accelerating rate but lack employees with the necessary capabilities to maximize the impact of gene

72、rative AI18.If we secure all assessments and lock out generative AI,we will fail to help students engage productively and responsibly with AI.Again,the pedagogically meaningful integration of generative AI into the curriculum,in service of learning disciplinary knowledge,skills,and dispositions,is k

73、ey here.Through this,we have an opportunity to build our students ability to engage with AI productively and responsibly.Another medium-term challenge is reaffirming the role of the university.Student engagement has been a growing issue,exacerbated by the COVID-19 pandemic,and the allostatic load fr

74、om a cumulation of stressors had made students question the purpose of higher education even before ChatGPT was released19.Since then,generative AI has become an alluring answer to activities which students may perceive as busywork20,which again raises questions around what we are asking students to

75、 do in higher education.The foresight workshop brought this into sharp focus,through a recognition that the current models of university are being disrupted through internal and external forces21.How to use this whitepaper to inform action This whitepaper aims to support institutions to move into th

76、e short-and medium-term future with generative AI by offering a set of practical elements that universities need to consider and put into action.As a point-in-time summary and direction-setting tool,the recommendations in this whitepaper will most likely need refreshing as higher education evolves a

77、longside the capabilities of generative AI.One underlying philosophy for this whitepaper is to reframe the approach to generative AI from policing to possibilities.With the increasing ubiquity of generative AI functionality in existing platforms,and availability of generative AI-17 For example,Scarf

78、e et al.(2024)A real-world test of artificial intelligence infiltration of a university examinations system:A“Turing Test”case study,https:/doi.org/10.1371/journal.pone.0305354 and Ibrahim et al.(2023)Perception,performance,and detectability of conversational artificial intelligence across 32 univer

79、sity courses,https:/doi.org/10.1038/s41598-023-38964-3 18 IDC(2024)The Business Opportunity of AI report.https:/ 19 McMurtie(2022)A stunning level of student disconnection.https:/ 20 McMurtie(2024)Cheating has become normal.https:/ 21 Joffres and Rey-Saturay(2024)The University at a Crossroads-Reima

80、gining Higher Education in an Age of Disruption.https:/www.apru.org/resources_report/generative-ai-in-higher-education-foresight-workshop/Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 12 specific tooling,and applications22,it is not feas

81、ible nor desirable to restrict,limit,or ban generative AI,nor to be overly fearful of what is left for humans.Rather,our approach is to consider what is now possible because generative AI is here.However,we are mindful of disciplinary and other contexts that necessarily mean we should also not uncri

82、tically embrace AI.To help universities approach this challenge head-on,this whitepaper identifies key phases of development and actions that can be taken by leaders,educators,researchers,and students within their contexts,considering their spheres of control,influence,and concern.These actions are

83、presented in this whitepaper as rubrics which describe,for each of these stakeholder groups,suggested levels of maturity from emerging,to established,to evolved,to extending.As with all rubrics,an individual,group,or institution may not neatly sit within one of these four levels.Development may also

84、 not be linear in all cases.Rather,the rubrics are a suggested starting point to position where one is currently operating,and what actions may be useful to consider as next steps.Acknowledgements The whitepaper has been informed by the virtual and in person workshops23 that have been conducted as p

85、art of this APRU project,as well as many other conversations around the topics that the authors have had in collaboration and as part of their own roles.We acknowledge the insightful feedback and further resources provided for this whitepaper by a number of leading thinkers and doers in generative A

86、I in the context of higher education that this project brought together:Alexandros Papaspyridis,General Manager,Nefos Consulting,Dubai Cecilia K.Y.Chan,Professor,Teaching and Learning Innovation Centre(TALIC)/Faculty of Education,The University of Hong Kong,Hong Kong Fun Siong Lim,Head of the Centre

87、 for Applications of Teaching&Learning Analytics for Students,Nanyang Technological University,Singapore 22 We take the definitions given in Microsofts Generative AI Tech Stack as outlined in the Australias Opportunity in the new AI economy report:foundation models are the“large generative AI models

88、 trained on vast datasets”;tooling refers to the framework and tools that go into generative AI applications,and applications being the“software solutions”used by end users such as students and educators.23 APRU(2023)The Future of Generative AI in Higher Education.https:/www.apru.org/our-work/univer

89、sity-leadership/generative-ai-in-education/Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 13 Michelle Banawan,Academic Program Director,Bachelor of Science in Data Science and Business Administration,Asian Institute of Management,Philippi

90、nes Ping Yein Lee,Adjunct Professor,UMeHealth Unit,Universiti Malaya,Malaysia Sean McMinn,Director of Center for Education Innovation,Hong Kong University of Science and Technology,Hong Kong Sergio Celis,Associate Professor,School of Engineering and Sciences,Universidad de Chile Stephen Aguilar,Asso

91、ciate Professor of Education,Associate Director,USC Center for Generative AI and Society,University of Southern California,United States Tim Fawns,Associate Professor,Monash Education Academy,Monash University,Australia For full transparency,generative AI applications were used in the development of

92、 this whitepaper in the following ways:NotebookLM assisted with source summarization and search.Claude assisted with proposing and critiquing descriptors for the CRAFT rubrics,summarizing and analyzing sources,and drafting the executive summary.Front cover and decorative images generated via openart

93、.ai using Flux(dev)model.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 14 Five areas for action Immediate key areas of activity There are three core areas of focus for universities to enable work towards the goal of productively and resp

94、onsibly integrating generative AI into their education,research,and operational functions.A combination of and balance between(1)rules,(2)access,and(3)familiarity is needed to enable appropriate adoption.A lack,or misbalance,of one or more of these areas may lead to ethical,privacy,security,academic

95、 integrity,or other challenges.These three areas are underpinned by a foundational layer of(4)trust between students,educators,leaders,vendors,partners(industry,government,and community),and AI itself.Rules,access,familiarity,and trust are then situated in,and influenced by,an institutions local,reg

96、ional,and even global(5)culture that includes attitudes,philosophies,and perspectives of individuals and groups of society,academia(universities and subunits),and governments.Together,these make the CRAFT framework(Figure 1)for generative AI adoption in higher education.We unpack each of these compo

97、nents,along with implications for different stakeholder groups along their generative AI journey.Figure 1.Interaction between the five core areas of activity needed to address generative AI in higher education.Association of Pacific Rim Universities Generative AI in higher education:Current practice

98、s and ways forward 15 Rules Establishing meaningful rules is critical to establishing the responsible use of generative AI and helps to build trust.These rules include principles,policies,guardrails,and guidelines that govern how individuals within an institution engage with generative AI,as well as

99、 how the institution approaches the technology.At a high level,creation of principles and position statements is one way that many APRU institutions approached their initial response to the technology and its implications,establishing high level rules for engagement with generative AI.One of the mos

100、t obvious and pressing reasons for having rules around generative AI centers on academic integrity and the veracity of higher education awards.That is,given generative AI can produce high quality student-like work25,how should assessments appropriately assure that learning has occurred?In other word

101、s,the focus on assessment should be around validity;that is,are we measuring a students actual capability26?This necessitates reconsideration of assessment regimes,because waiting,or perhaps hoping,for tools to detect writing authored by generative AI models with a sufficiently high degree of accura

102、cy and reliability is not the answer.AI detection tools yield uncomfortable levels of 24 Lodge et al.(2023)Assessment reform for the age of artificial intelligence.https:/www.teqsa.gov.au/guides-resources/resources/corporate-publications/assessment-reform-age-artificial-intelligence 25 For example,S

103、carfe et al.(2024)A real-world test of artificial intelligence infiltration of a university examinations system:A“Turing Test”case study,https:/doi.org/10.1371/journal.pone.0305354,and Borges et al.(2024)Could ChatGPT get an engineering degree?Evaluating higher education vulnerability to AI assistan

104、ts,https:/doi.org/10.1073/pnas.2414955121.26 Dawson et al.(2024)Validity matters more than cheating.https:/ Case studies Philippines Most higher education institutions in the Philippes quickly established rules around generative AI usage by faculty and students through acceptable use policies.The Un

105、iversity of the Philippines released principles-based guidelines that balanced positive use with negative impacts,focusing on beneficence,human agency,fairness,safety,environmental sustainability,and more.Australia Australias Tertiary Education Quality and Standards Agency,the federal regulator of A

106、ustralian universities,has produced a document,Assessment reform for the age of artificial intelligence24,which outlines two key principles:1.Assessment and learning experiences equip students to participate ethically and actively in a society where AI is ubiquitous.2.Forming trustworthy judgements

107、about student learning in a time of AI requires multiple,inclusive and contextualized approaches to assessment These principles encourage institutions to simultaneously engage in integrating generative AI into assessment and learning,whilst assuring that learning has occurred through trustworthy ass

108、essment positioned at meaningful points along a students journey.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 16 false positives and false negatives and is also easily defeated with creative prompting or purpose-built AI humanizer tools

109、27.Between higher education institutions,this will look different depending on the academic context,but the program level will be the natural location to distribute assessments that can assure students have attained learning outcomes,as well as define how,when,and if generative AI can be used in sup

110、port of learning.More broadly,the values of academic integrity(including fairness,honesty,respect,and responsibility)are highly compatible with the legitimate use of generative AI for learning.Having the right rules at an institutional level can therefore catalyze a productive engagement with genera

111、tive AI,beyond the narrow perspective of seeing AI as just cheating.This can help to reduce the worry that is currently pervasive around generative AI use31.Another reason for establishing clear institutional rules revolves around data privacy,intellectual property,and security.There have been well-

112、known cases where private or confidential information has been inadvertently provided to AI vendors,potentially for training future AI models,as part of unprotected conversations.This often occurs due to a lack of clear guidelines or insufficient awareness of rules around appropriate use.These rules

113、 would include the safe use of these applications,such as what data can be provided to them,which are safe to use,27 For example,Elkhatat et al.(2023)Evaluating the efficacy of AI content detection tools in differentiating between human and AI-generated text,https:/ Weber-Wulff et al.(2023)Testing o

114、f detection tools for AI-generated text,https:/doi.org/10.1007/s40979-023-00146-z 28 TEQSA(2024)Gen AI strategies for Australian higher education:Emerging practice.https:/www.teqsa.gov.au/guides-resources/resources/corporate-publications/gen-ai-strategies-australian-higher-education-emerging-practic

115、e 29 Liu&Bridgeman(2024)Frequently asked questions about the two-lane approach to assessment in the age of AI.https:/educational-innovation.sydney.edu.au/teachingsydney/frequently-asked-questions-about-the-two-lane-approach-to-assessment-in-the-age-of-ai/30 Liu(2024)Menus,not traffic lights:A differ

116、ent way to think about AI and assessments.https:/educational-innovation.sydney.edu.au/teachingsydney/menus-not-traffic-lights-a-different-way-to-think-about-ai-and-assessments/31 Students perspectives on AI in higher education.https:/aiinhe.org/wp-content/uploads/2024/10/aiinhe_surveyinsights.pdf Ca

117、se study Tertiary Education Quality and Standards Agency(TEQSA),Australia In mid-2024,202 Australian higher education providers responded to a request for information from the countrys higher education regulator,TEQSA,sharing institutional approaches to the risks posed by generative AI to the veraci

118、ty of awards.TEQSA curated key emerging practices in a practical toolkit28,focusing on process,people,and practice.In the toolkit,TEQSA highlighted key practices around assessment security and academic integrity,emphasizing the importance of assessing process,assessment validity,and a program-level

119、approach to assessment.A key transformational practice identified by TEQSA in mitigating assessment risk was the two-lane approach to assessment redesign29,where lane 1 supervised assessments were used for the assessment of learning,and lane 2 open assessments were used as assessment for learning.Th

120、e use of AI is scaffolded and supported in lane 2,applying a menu of typologies30 of generative AI use.This was highlighted because a menu analogy emphasizes choice and suitability,as opposed to a traffic light or assessment scale approach which suggests that one can restrict or control AI use(one c

121、annot),or that there is a linear gradation of AI use(there is not).Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 17 and the contexts and configurations of their use(e.g.what data goes back to the AI vendor to refine or optimize AI models

122、).For example,unpublished research findings may be considered too sensitive to share with certain AI applications,including some hosted in the cloud.Cloud-based platforms potentially present risks from unauthorized access and exposing data to these services could compromise security and ownership of

123、 research data.From an education perspective,do educators have the right to upload student work to AI tools without informed consent for the purposes of generating feedback?Establishing rules,considering the pace at which generative AI progresses,and in the face of its ubiquity and ease-of-access,br

124、ings certain challenges.It is important for rules to be as forward-looking as possible and to be revisited regularly as the technology changes32,becomes more widespread and integrated into existing platforms33,and as the culture around generative AI adapts.For example,implementing rules around AI-pr

125、oofing assessments is not forward-looking because AI capabilities will likely advance faster than educators can redesign assessment tasks.As generative AI impacts various disciplines in different ways,it is also important for rules to allow for disciplinary nuances and interpretation34 while also co

126、nsidering and encouraging interdisciplinarity.In many ways,the wide accessibility and use of generative AI-enabled applications amongst the university population has meant that rules have already fallen behind in many institutions,which makes it more difficult to take advantage of benefits and mitig

127、ate risks35.As the most avid(albeit not necessarily productive,sophisticated,or responsible)current users of AI36,students should be central to any discussions around rules.This is an opportunity to engage students as partners and co-creators in defining and applying approaches:as a cohort group,the

128、y are engaged,eager for guidance,and generally aware of how important proficiency with generative AI is going to be as they move through and beyond their time at university.The extensive literature of students as partners as an approach for course design and evaluation offers practical guidance on h

129、ow to approach this37.Looking to the future,it is critical that rules are designed for a future state where AI is increasingly capable and integrated in many digital tools and new,as-yet unknown,possibilities.Rule design also needs to help catalyze and guide a shift towards responsible human-AI coll

130、aboration.To this end,the following rubric(Table 1)can be used to help situate your institutional and local progress and consider key action areas for development.32 Joffres and Rey-Saturay(2024)Generative AI in Higher Education Sensemaking Workshop Proceedings.https:/www.apru.org/resources_report/g

131、enerative-ai-in-higher-education-sensemaking-workshop/33 Justus&Janos(2024)Your AI Policy Is Already Obsolete.https:/ 34 Joffres and Rey-Saturay(2024)Generative AI in Higher Education Creative Sandbox Report:Prototype Concepts for Higher Education in the AI Future.https:/www.apru.org/resources_repor

132、t/generative-ai-in-higher-education-creative-sandbox/35 Australian Government(2024)Study Buddy or Influencer.https:/www.aph.gov.au/Parliamentary_Business/Committees/House/Employment_Education_and_Training/AIineducation/Report 36 Digital Education Council(2024)Global AI Student Survey.https:/ 37 Heal

133、ey et al.(2016)Students as partners:Reflections on a conceptual model.https:/doi.org/10.20343/teachlearninqu.4.2.3 Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 18 Rules:Self-positioning rubric Table 1.Rubric for establishing rules aroun

134、d engaging with generative AI.Emerging Established Evolved Extending Leaders Desire for/initial discussions leading to drafts of institution-wide principles and policies,such as privacy,security,ethics,and integrity.Formation of some governance structures.Committees and working groups formed,leading

135、 to principles and policies around privacy,security,ethics,compliance,quality assurance,and academic integrity as relates to generative AI.AI governance structure with clear accountability.Clear guidance and resources provided and communicated to educators,researchers,and students.Impacts on diversi

136、ty,equity,and inclusion are considered.Collaboration internally and externally(other universities,industry,accrediting bodies)on standards and resources.Regular validation and review of rules.Comprehensive AI strategy,monitoring,and quality assurance mechanisms articulated and integrated into instit

137、utional plans.Diversity,equity,and inclusion are central to institutional approaches to AI.Cross-sector partnerships(with industry,accrediting bodies,government,community)to define responsible AI use.Influencing wider policies such as industry practices and codes of conduct.Educators Uncertainty abo

138、ut permissible roles for AI in teaching,learning,and assessment.Ad hoc rules set by individual educators.Some acknowledgement of AI use(or not)in course documents.May be banning AI entirely in assessments.Institutional rules about AI in teaching,learning,and assessment are clearly understood,consist

139、ently cascaded and appropriately applied in different disciplinary contexts.Responding to the need to assure learning outcomes and prepare students for the future.Providing feedback on policy effectiveness for on-going enhancement.Aligning course-specific nuances of institutional rules to disciplina

140、ry needs.Securing assurance of learning outcomes at key points of students journeys.Consideration and integration of AI in curriculum review processes.Contributing to educator-led AI working groups to influence policy directions and wider practice.Researchers Ad hoc use with limited institutional gu

141、idance.May be unclear about data security requirements.Developing discipline-specific guidelines and approach for responsible AI use in research.Safely using AI in research,maintaining data security.Involving research ethics boards in generative AI decisions.Active contributions to refining institut

142、ional rules on AI for research.Contributing to AI research standards and developing best practices for specific domains.Collaborating on AI-enabled research methodologies.Contributing to global AI research standards.Students Basic awareness of rules and policies around AI use,but some apprehension a

143、bout application in different learning contexts.Clear understanding of permissible AI use in learning and assessment and adherence to different guidelines across courses and programs.Active engagement in discourse around AI.Student partnership in AI governance.Student-led initiatives to ideate,refin

144、e and feed back on AI policies.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 19 Access Equitable availability of generative AI applications for students,educators,and leaders across the institution is essential.This may include licenses

145、to discipline-specific applications within certain departments,general purpose AI platforms available across an institution,and ensuring presence of supporting infrastructure.Inequitable access to such a critical technology as generative AI risks exacerbating existing digital divides,opening new rif

146、ts,and entrenching disadvantage across the system38.Foundational to this is having access to enabling infrastructure such as internet connections and computing devices,which remains particularly challenging for marginalized and low-and middle-income communities and even countries39.The cost of acces

147、sing AI platforms and subscriptions can be prohibitive for many individuals,institutions,and whole jurisdictions,potentially creating a new form of digital inequity where access to advanced AI capabilities is determined by financial resources.For example,paid subscriptions to latest-model generative

148、 AI applications usually cost between USD20-30 per month,per platform;paid access typically grants more reliable access to frontier models,improved functionality such as data analysis,and enhanced output quality.It is important that institutions,governments,and AI vendors work together to supply AI

149、applications and tooling to ensure that essential AI functionalities are available free of charge to students,educators,and researchers40.This may involve institutional or governmental agreements with vendors,or the deployment and use of open-weights AI models.Considerations of equitable access also

150、 include potential barriers relating to disability,culture,and language.AI vendors have a responsibility to ensure AI interfaces are designed with accessibility in mind,and that students with disabilities receive the support and accommodations they need in using AI effectively41.It must also be reco

151、gnized that generative AI may be a powerful assistive technology for certain students,such as helping neurodivergent students to organize and reprocess material.Another aspect of equity is related to the predominantly Western perspectives in training datasets that may perpetuate biases and limit the

152、 relevance of AI applications for learners from diverse backgrounds,or even limit the capabilities of AI models in 38 NSW Parliament(2024)Artificial intelligence(AI)in New South Wales.https:/www.parliament.nsw.gov.au/committees/inquiries/Pages/inquiry-details.aspx?pk=2968 39 Australian Government(20

153、24)Study Buddy or Influencer.https:/www.aph.gov.au/Parliamentary_Business/Committees/House/Employment_Education_and_Training/AIineducation/Report 40 For example,Microsoft enabling access to OpenAIs frontier models for education in Hong Kong.41 Davis(2024)Developing Institutional Level AI Policies an

154、d Practices:A Framework.https:/wcet.wiche.edu/frontiers/2023/12/07/developing-institutional-level-ai-policies-and-practices-a-framework/Case study The Philippine governments Department of Science and Technology has partnered with the National University and Bicol University to make available an AI a

155、pplication that helps people engage with databases using natural language queries.The initiative is designed to encourage adoption of this AI in universities,colleges,local government,and by the general public.The government sees this as a way to make information more accessible to citizens,especial

156、ly those who may not be familiar with English.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 20 certain languages.Models trained on a corpus of material in a specific language other than English may emerge for specific geographies or purp

157、oses,such as to avoid neglecting certain cultures and languages42.Some stakeholders may also have valid and deeply held convictions about the ethics of generative AI systems and elect to limit their own access.For example,non-users of AI may hold concerns about the environmental impact of AI inferen

158、ce,and the ethical labor practices of AI companies in preparing models43.These factors should be considered when institutions are making decisions about generative AI applications and may lead to the selection of more ethical or sustainable options or giving individuals the agency to conscientiously

159、 object whilst providing equitable alternatives.Another consideration relates to the increasing ubiquity of generative AI functionality in existing platforms.For example,some publishers are adding generative AI summaries to existing scholarly databases used by researchers and students.In these cases

160、,access may be automatic,in which case institutions will need to invoke other elements of the CRAFT framework to appropriately respond,such as rules to ensure data protections are in place,and familiarity to ensure users are aware of the opportunities and limitations of generative AI.Looking to the

161、future,it is critical that institutions seek to provide access to state-of-the-art AI tooling and applications to ensure their students,educators,and leaders can learn how to use AI productively and responsibly.This may require more flexible licensing arrangements with vendors to permit responsivene

162、ss to new developments.The following rubric(Table 2)can be used to help situate your institutional and local progress and determine key action areas for development.42 Biever(2024).Chinas ChatGPT:Why China is building its own AI chatbots.https:/ 43 McDonald et al.(2024)Apostles,Agnostics and Atheist

163、s:Engagement with Generative AI by Australian University Staff.https:/eprints.qut.edu.au/252079/Case studies Universities are leveraging Microsofts Azure OpenAI services through custom-built platforms designed for a higher education context.These initiatives provide equitable access to state-of-the-

164、art AI models for all stakeholders at an institution,as access to the underlying AI tooling is provisioned by the institution.Tecnolgico de Monterrey Tecnolgico de Monterrey have developed TECgpt,a generative AI ecosystem based on Microsofts Azure OpenAI service.TECgpt makes available to the communi

165、ty number of different components,including ChatGPT-like functionality,and language processing capability on top of the institutions own knowledge bases.Students can ask TECbot tutors for help,approach it for administrative advice,and teachers can use it to create teaching material.The University of

166、 Sydney The University of Sydney has developed the Cogniti platform,to allow educators to create their own AI agents,with integration into the learning management system.Also built on Microsofts Azure OpenAI service,educators can control the behaviour and knowledge base of their AI agents,understand

167、 how students interact with it,and share their agents with others.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 21 Access:Self-positioning rubric Table 2.Rubric for providing equitable access to generative AI technologies.Emerging Establ

168、ished Evolved Extending Leaders Identifying a need for different resources(technology,people)for investment.Initiating discussions with potential AI vendors and/or local development teams.Budgets identified and allocated to AI resources.Alignment of procurement to established rules.Pilot projects ar

169、e supported.Small-scale availability of key AI applications.Consideration of accessibility and inclusion issues in available AI platforms.Some evaluation of AI application efficacy.Institution-wide financially sustainable availability of AI applications using frontier models.Discipline-specific appl

170、ications widely and equitably available.New resources considered as part of annual planning.Consideration of ethical AI models and tooling.Inter-institutional collaboration to secure cost-effective,equitable access to AI tooling and applications.Systematic evaluation of AI application efficacy.Colla

171、boratively developing novel AI applications in partnership with other institutions and AI vendors,such as through innovation hubs.Creating AI innovation hubs in collaboration with community partners and stakeholders.Educators Limited and/or hesitant exploration of AI applications or functionality re

172、levant to learning,teaching,and assessment.Free tools used.A variety of AI applications are utilized,built into learning design for courses as appropriate.Encourages students to select and use AI applications.Working with IT to ensure classroom infrastructure supports AI use.Discipline-specific AI a

173、pplications are embraced in collaboration with leadership.Encourages students to leverage free access to relevant AI applications.Participating in decisions and evaluations about AI application availability and effectiveness.Co-designing and building their own educational tools using self-serve AI a

174、pplications.Leading inter-institutional collaborations on development of AI applications.Advising on AI use for specific applications.Researchers Use of free or commercially available AI applications with limited data protection.Using institution-provided AI applications for research.Piloting other

175、discipline-specific AI tooling or applications to accelerate research activities.Involved in selection,deployment,evaluation,and cross-disciplinary sharing of research-enabling AI applications or tooling(e.g.data analysis,literature review,code generation).Collaborating on building AI-enabled resear

176、ch applications and tooling.Integration of AI tooling with research infrastructure.Students Limited awareness and use of available AI tools.Reliance on free,mass-market AI applications.Accessing institution-provided AI applications.May use other AI-enabled applications to support personal learning a

177、nd research in curricular,co-curricular,or extra-curricular contexts.Actively involved in requirements for,selection of and deployment of AI applications,possibly specific to the discipline.Have equitable access to infrastructure to leverage AI applications and tooling.Co-designing AI applications a

178、nd uses with educators.Access to advanced AI applications and tooling used for research and industry.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 22 Familiarity This represents how well students,faculty,and staff understand and are comf

179、ortable with the ways they may use generative AI for their day-to-day work related to the institution.We have purposely used the word familiarity here instead of skill,as not all stakeholders will(or will need to)develop advanced skill with using generative AI.However,all stakeholders need to have f

180、oundational knowledge of the possibilities AI affords,where it could and should be used,and have the ability to apply AI to their everyday activities44.Familiarity also emphasizes an awareness of the broader context of generative AI parallel to its application,including ethics,privacy,and safety45.T

181、he imperative to develop familiarity with generative AI is rooted in universities needing to design and delivery coursework and research activities that prepare students for their future.However,recent reports suggest that universities are not providing the necessary familiarity-building activities

182、that students need46.A large contributing factor is that educators,researchers,and leaders themselves are struggling to build their own familiarity,often because their institution lacks a generative AI strategy47.This is despite staff training and scope for experimentation being some of the most sou

183、ght-after developments to help meet AI literacy needs48.However,building staff and student familiarity needs to be contextualized within the cultural environment of higher education,including perspectives on the place of AI within higher learning(see later section on culture).Further,keeping up to d

184、ate with the rapid pace of generative AI developments 44 Brodnitz(2024)A New Framework for AI Upskilling Across Your Organization.https:/ 45 World Economic Forum(2024)Shaping the Future of Learning:The Role of AI in Education 4.0.https:/www3.weforum.org/docs/WEF_Shaping_the_Future_of_Learning_2024.p

185、df 46 Digital Education Council(2024)Global AI Student Survey 2024.https:/ 47 Microsoft(2024)AI in Education:A Microsoft Special Report.http:/aka.ms/AIinEDUReport 48 McDonald et al.(2024)Apostles,Agnostics and Atheists:Engagement with Generative AI by Australian University Staff.https:/eprints.qut.e

186、du.au/252079/Case studies Nanyang Technological University Nanyang Technological University in Singapore is developing a university strategy for an ecosystem of responsible AI applications for teaching and learning,to address governance and responsible use of AI,promote AI literacy and enable experi

187、mentation through a local sandbox environment.This central institutional approach is promoting use of common language,common measures of impact and responsible use,and a common platform.Asian Institute of Management At the Asian Institute of Management(AIM)in the Philippines,generative AI is thought

188、fully incorporated into teaching,learning,and assessment practices to enhance student outcomes while maintaining academic integrity.By allowing students to use AI for brainstorming,initial drafts,and study guide creation,AIM provides practical AI experience while ensuring that critical tasks like ca

189、se analysis and reflections remain authentically student driven.This balanced approach not only familiarizes students and faculty with AI tools but also reinforces AIMs commitment to pedagogically meaningful use of GAI,supporting ethical and impactful learning experiences.Association of Pacific Rim

190、Universities Generative AI in higher education:Current practices and ways forward 23 is increasingly difficult.Several universities in the Pacific Rim are starting to establish centers for AI that variably meet the practical and/or research needs of the institution.That said,familiarity-building ini

191、tiatives for faculty are still generally nascent or piecemeal,even though these are emerging as a necessary precondition for wider adoption49.Familiarity with generative AI in higher education also necessarily includes how it can be incorporated into teaching,learning,and assessment practices and de

192、signs in pedagogically meaningful ways50.For example,it may be more effective to provide students with intentionally designed generative AI applications that are aware of common misconceptions,promote problem solving,and develop metacognitive skills51 a,than to provide students with unguided,general

193、-purpose generative AI that may help answer questions but turn out to be a crutch52 that can allow students to avoid important cognitive labor and on which they can become over-reliant with potentially adverse long-term(and as-yet unknown)impacts.Overall,we need to take a“pedagogy first”approach to

194、ensure that student learning needs and educators pedagogical intent are foregrounded,along with the deeply relational nature of teaching and learning53.From a student perspective,these tools are alluring.They can make the things a student needs to do to satisfy assessment requirements rapid and fric

195、tionless.A deliberately extreme example,which many could easily imagine when ChatGPTs capabilities first became known54,is a student using the tool to write an entire essay or assignment.This would deprive the student of the desirable cognitive effort needed to learn from undertaking the assignment

196、and building competencies that are important for the course or program of study.Writing is a process closely tied to thinking,which is not a process we want to see our students short-circuit entirely.Students need to develop a nuanced view,supported by their educators,of not only how to use generati

197、ve AI to support learning,but when not to rely on them.Students also need to develop strong metacognitive processes such as self-regulated 49 Joffres and Rey-Saturay(2024)Generative AI in Higher Education Sensemaking Workshop Proceedings.https:/www.apru.org/resources_report/generative-ai-in-higher-e

198、ducation-sensemaking-workshop/50 Microsoft(2024)AI in Education:A Microsoft Special Report.http:/aka.ms/AIinEDUReport 51 For example,Lai et al.(2024)Leveraging Process-Action Epistemic Network Analysis to Illuminate Student Self-Regulated Learning with a Socratic Chatbot.https:/doi.org/10.35542/osf.

199、io/b9vq6 52 Bastani et al.(2024)Generative AI Can Harm Learning.https:/ 53 Joffres and Rey-Saturay(2024)Generative AI in Higher Education Sensemaking Workshop Proceedings.https:/www.apru.org/resources_report/generative-ai-in-higher-education-sensemaking-workshop/54 See,for example,Marche(2022)The Co

200、llege Essay Is Dead.https:/ study At the Hong Kong University of Science and Technology,the Centre for Education Innovation are trialing ChatGPT as a design assistant in educational course design.AI is used to help align course learning outcomes to assessment design,following Blooms taxonomy.The AI

201、speeds up processes such as mapping cognitive processes to knowledge dimensions,with the educator guiding this process.This allows educators to be more reflective and thorough,augmenting human capabilities.Association of Pacific Rim Universities Generative AI in higher education:Current practices an

202、d ways forward 24 learning that promote autonomy,adaptability,and reflexivity,which will also help them to critically engage with generative AI55.For all stakeholders,AI ethics is also a critical element of familiarity.The UNESCO AI competency frameworks for teachers57 and students58 highlight aware

203、ness of debates around the ethics of AI as a key aspect,including the impact of AI on equity,environment,social justice,and human rights.A foundational familiarity with AI ethics will help inform how we use AI,what we do with AI outputs,which AI models and applications we use,how to consider potenti

204、al harms,and how we engage with vulnerable groups around AI.For example,awareness of the bias in training data and AI outputs may help students to be more careful about evaluating the perspectives or representations that AI presents.Awareness of the environmental impact of generative AI may lead res

205、earchers to choose simpler AI models for tasks like bulk summarization that do not require advanced reasoning capabilities.The UNESCO global AI ethics and governance observatory59 is a key resource for further investigation into international approaches to AI ethics,including UNESCOs own principles

206、which center on human rights and dignity,justice,diversity and inclusion,and environmental flourishing60.55 Lodge et al.(2023)Learning with Generative Artificial Intelligence Within a Network of Co-Regulation.https:/doi.org/10.53761/1.20.7.02 56 FLO MicroCourse:Future Facing Assessments OER(2023)htt

207、ps:/scope.bccampus.ca/course/view.php?id=619 57 Miao&Cukurova(2024)AI competency framework for teachers.https:/unesdoc.unesco.org/ark:/48223/pf0000391104 58 Miao&Shiohira(2024)AI competency framework for students.https:/unesdoc.unesco.org/ark:/48223/pf0000391105 59 Global AI Ethics and Governance Ob

208、servatory.https:/www.unesco.org/ethics-ai/en 60 Ethics of Artificial Intelligence.https:/www.unesco.org/en/artificial-intelligence/recommendation-ethics Case studies Familiarity is an area where collaboration and open sharing and licensing of resources to facilitate reuse and adaption to local conte

209、xts is beneficial.The University of British Columbia Work undertaken at the University of British Columbia,Canada,done on behalf of BC Campus,a provincial organization that supports all post-secondary education institutions in British Columbia,Canada,has led to the creation of a free,openly licensed

210、 faculty development course on the design of assessments that invite student use of generative AI56.University of Southern California By leveraging interdisciplinary collaboration between its Rossier School of Education and the Institute for Creative Technologies,the University of Southern Californi

211、a has created programs like the Generative AI Fellows,which empower students to explore and critically evaluate AIs potential and ethical implications in education.This has led instructors and students alike to explore generative AI technologies in a supportive environment.The University of Sydney S

212、tudent partners at the University of Sydney have developed a public-available,Creative Commons-licensed resource site to help students and educators use AI productively and responsibly.The AI in Education site houses information on what generative AI is,the integrity and other ethical considerations

213、 of its use,and many real-world examples of AI prompts that students themselves find useful to support learning,assessment,and career growth.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 25 Wider contextual familiarity with how universit

214、y partners are engaging with generative AI is also important.For example,the ways that industry are engaging with AI will impact on how AI is incorporated into higher education research and teaching practices.Conversely,university researchers and educators,as disciplinary experts,play a key role in

215、influencing and leading how industry,government,and the community productively and responsibly engage with generative AI.Additionally,reactions of the community to generative AI,such as around ethics and safety,must inform how students and researchers build their AI literacies.Looking to the future,

216、there are many opportunities for building familiarity with various university stakeholders.Students-as-partners initiatives can play a powerful role in normalizing,sharing,and celebrating productive and responsible applications of generative AI.These may include novel pedagogies which are afforded b

217、y these technologies,such as the scaling of personalized simulation environments,or new forms of discussion and collaboration enabled by AI conversation partners.Some have called for the development of pedagogical intelligence to engage with AI in education in new ways61.The interplay between univer

218、sity stakeholders also underscores that familiarity is not just an individual feature but a collective,organizational one if an institution has members with varying but complementary levels of familiarity with generative AI,collectively the familiarity of the organization is established as long as t

219、here is alignment and collaboration.For example,if faculty members understand the disciplinary applications of AI,and instructional designers understand the affordances of AI in pedagogy,and educational technologists understand the capabilities of different AI applications,their combined AI familiar

220、ity can be powerfully applied.Many universities have set up internal communities of practice where educators and researchers can exchange ideas,learn about key updates,and build familiarity together.Use the following rubric(Table 3)to help situate your institutional and local progress and determine

221、key action areas for development.61 Daz and Nussbaum(2024)Artificial intelligence for teaching and learning in schools:The need for pedagogical intelligence.https:/doi.org/10.1016/pedu.2024.105071 Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways fo

222、rward 26 Familiarity:Self-positioning rubric Table 3.Rubric for building familiarity with generative AI across an institution.Emerging Established Evolved Extending Leaders Growing awareness of AI and early development of AI literacies.Focus on risks and their mitigation.No or ad hoc resourcing arou

223、nd training.Limited personal experience with AI.Well-informed and confident about AI capabilities and ethical considerations.Advocacy for integrating AI into some aspects of institutional work.Resourcing groups to train and work with people to use AI.Occasional or periodic users of AI.Establishing r

224、esource hubs or training modules to inform responsible and productive AI use.Well-developed fluency with AI including opportunities and risks.Fostering a culture of experimentation,opportunity,and investment across the institution.Inspiring groups to explore and share openly.Implementing ethical app

225、roaches to AI use.Regular users of AI.Evaluating efficacy of training.Anticipate and prepare the institution for future AI developments.Developing long-term strategies for AI integration in collaboration with other institutions and industry,government,and community.Educators Curiosity about AI and e

226、ngaging with workshops or resources to build basic understanding.Permitting students to use AI for learning in some course contexts.Exploring basic AI ethics concepts.Limited integration with learning design of courses Comfortable using AI in different ways in teaching and assessment.Utilizes resour

227、ces to support student engagement with AI.Encourages students to use AI in learning and assessment.Integrating AI ethics considerations into courses.Appropriate integration into learning design of own courses Deep familiarity with AI and continual engagement in updating knowledge.Actively and openly

228、 sharing with peers and students.Engaging with students as partners in learning about and using AI.Integrating tools into learning design within and possibly beyond own discipline.Engaging with professional bodies to become familiar with industry applications of AI to inform teaching.Developing new

229、pedagogical approaches that integrate AI into learning design and activities.Preparing curriculum to meet the needs of an AI-infused world.Leading and influencing other educators in applying AI creatively,productively,and responsibly.Researchers Initial experimentation with AI applications for resea

230、rch tasks.Attending workshops or sessions to build basic AI literacy.Able to evaluate AI applications and tooling for research appropriateness.Peer discussions about AI use in research methods.Developed expertise in AI applications within research domain.Leading discussions and mentoring on AI use i

231、n research.Actively contributing to AI methodology development.Developing approaches for ethical AI use in research.Pioneering new AI applications in research.Leading cross-disciplinary initiatives in AI research uses.Association of Pacific Rim Universities Generative AI in higher education:Current

232、practices and ways forward 27 Emerging Established Evolved Extending Students Basic or unsophisticated use of AI,in ways guided by educators,peers,or other influences.Use may be predominantly for providing answers/looking things up rather than scaffolding learning.Routine,productive use of AI to sup

233、port learning,not replace cognitive effort.Sound understanding of AI benefits and limitations,and critical evaluation of AI output.Appreciation of AI ethics.Able to critically evaluate the application of AI across different domains,in the context of their own learning processes.Skilled at integratin

234、g AI across various aspects of academic life,starting to work in partnership with AI.Engaging in peer-to-peer learning about AI.Contributing to AI ethics debates.Partnering with the institution to boost familiarity across the student body.Student-initiated projects around AI use in education that be

235、nefit community.Exploring AIs potential impact on future careers.Developing deeper collaborative ways of working with AI.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 28 Trust Trust is a key element in helping people adopt AI technologie

236、s.Users trust can be conceptualized as between the user and the technology,and the user and the vendor,being influenced by cognitive,emotive,and behavioral dimensions62.However,the trust element in the CRAFT framework extends beyond the relationship between people,AI,and vendors.There are many other

237、 trust pairs that are important to consider,such as between students and educators,between educators and leaders,between universities and vendors,between researchers and the community,and more.There are negative consequences when trust is eroded between key trust pairs(Table 4).Table 4.Some potentia

238、l consequences when trust is eroded between selected trust pairs in the context of generative AI and universities.Trusting party Party being trusted Consequence of trust erosion Students Educators Feelings of hypocrisy and unfairness Educators Students Suspicion,descension into adversarial mindsets,

239、reliance on AI detection Leaders Educators Managerialism,overbearing rules,removal of access,discouraging experimentation Educators Leaders Fear of retribution,lack of experimentation Students AI Fear and avoidance Educators AI Fear,avoidance,and negative advocacy Community Researchers Disbelief in

240、research outcomes University Vendors Overbearing procurement processes,lack of engagement and access Community University Doubting the validity of awards,doubting the value of a university education to prepare graduates One key trust pair exists between students and educators.Students are recognizin

241、g that AI is ubiquitous and would use it even if they are instructed not to in increasingly larger proportions63.When coupled with educators generally being behind their students in engaging with generative AI64,and concerns around academic integrity and effects on learning,it is understandable that

242、 there is a rapidly widening trust gap between students and educators65.Mistrust is further exacerbated through use of surveillance and detection technologies66 that ostensibly aim to establish whether students have completed their own work but can be invasive,inaccurate,and easily 62 Yang and Wibow

243、o(2022)User trust in artificial intelligence:A comprehensive conceptual framework.https:/doi.org/10.1007/s12525-022-00592-6 63 Tyton Partners(2024)Time for Class 2024.https:/ Ibid.65 Coldwell(2024)I received a first but it felt tainted and undeserved:inside the university AI cheating crisis.https:/

244、66 Ross and McLeod(2018)Surveillance,(dis)trust and teaching with plagiarism detection technology.https:/doi.org/10.54337/nlc.v11.8760 Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 29 defeated67.If educators use AI to grade student work(

245、for example,to save time),the trust relationship is further impacted through perceptions of hypocrisy,inaccuracy,and unfairness68.To start to address educator-student trust around AI,educators could model brave transparency around their own use of generative AI,and work with students to develop rule

246、s(in the form of local expectations through to institutional policies)and build familiarity together,as these opportunities for partnership are currently not being met69.An important contributing factor is educators and researchers lack of trust in generative AI itself.There are many valid reasons f

247、or this,including distrust in its accuracy and reliability,concerns around diminishing human value and creativity,lack of respect for data sovereignty,and feelings of intimidation around the unknown70.While some of this may be mitigated through building familiarity with and demystifying generative A

248、I,there are some concerns around morality,professional ethics,and human exceptionalism are more deeply rooted(see later section on culture).Trust between educators and AI might be fostered by increasing the level of control and visibility of AI use by students human control and agency are seen as ke

249、y elements in enhancing trust in AI systems71.However,while having students conversations with AI visible to their educators may help build educator trust,it may erode student trust and needs to be framed with student learning and care at the center.More generally in the population,there are prevail

250、ing concerns around data privacy and security,safety,and transparency72.Uptake of AI also differs significantly between institutions.Part of the reason relates to the risk maturity and appetites of different universities,which intersects with the trust relationship between educators/researchers and

251、leaders.Engagement with a new general-purpose technology like AI benefits from experimentation and invention73,and educators/researchers need an environment,built by leaders,within which to feel safe to pilot and fail.Supporting safe experimentation,collegial sharing,and open dialogue are key action

252、s that institutional leaders can undertake in establishing an environment of trust.This is underpinned by a strong vision around productive and 67 For example,Perkins et al.(2024)GenAI Detection Tools,Adversarial Techniques and Implications for Inclusivity in Higher Education.https:/doi.org/10.48550

253、/arXiv.2403.19148 68 Digital Education Council(2024)Digital Education Council Global AI Student Survey 2024.https:/ 69 Ibid.70 McDonald et al.(2024)Apostles,Agnostics and Atheists:Engagement with Generative AI by Australian University Staff.https:/eprints.qut.edu.au/252079/71 Gillespie et al.(2023)T

254、rust in Artificial Intelligence:Global Insights 2023.https:/ 72 Ibid.73 Crafts(2021)Artificial intelligence as a general-purpose technology:an historical perspective.https:/doi.org/10.1093/oxrep/grab012 Case study The Chinese University of Hong Kong has developed the TellUs AI interview training pla

255、tform that is designed to help students and recent graduates prepare for interviews.It provides a mock interview experience and has been specifically designed to help interviewees practice answer coherence and relevance,as well as speech patterns and body language.Having been designed with these edu

256、cational goals in mind,students can trust the feedback provided by the platform and the platform itself.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 30 responsible use of generative AI in all aspects of a universitys work and buy-in fos

257、tered through co-design and shared decision-making.AI vendors like Microsoft,OpenAI,Anthropic,and Google play a key role as well.The trust relationship between universities and AI vendors is crucial to foster responsible and ethical use of AI.Given the hunger for training data by AI companies,there

258、are valid fears around the security and privacy of data provided to generative AI systems.Providing commercial data protection arrangements and mechanisms for users to opt out of data collection(or,better yet,have agency to opt in)are imperative to building this trust relationship,such as that affor

259、ded by Microsoft Copilots enterprise data protection arrangement.Recent developments such as AI nutrition labels74 and concerns over use of copyright material for AI training have helped to raise awareness amongst AI users and provide necessary visibility and interpretability around AI privacy issue

260、s.A 2023 analysis75 suggested that trust is central to AI adoption and there are four pathways to building trust in AI generally in the working population:(i)regulations and laws to make AI safe;(ii)realizing benefits of AI,(iii)addressing concerns about AI risks,and(iv)increasing understanding of,a

261、nd capability with,AI.Applied to the higher education context and within the CRAFT framework,these regulations correspond to rules,realizing benefits requires access and familiarity,while addressing concerns about risks and increasing understanding and capability correspond to familiarity.That is,tr

262、ust can be built by having rules that establish responsible use of AI,and by ensuring that students,educators,researchers,and leaders are able to understand and benefit from AI by using it in productive and ethical ways.Consider the following rubric(Table 5)to help situate your institutional and loc

263、al progress and determine key action areas for development.74 For example,https:/nutrition-facts.ai/or https:/openethics.ai/label/75 Gillespie et al.(2023)Trust in Artificial Intelligence:Global Insights 2023.https:/ Association of Pacific Rim Universities Generative AI in higher education:Current p

264、ractices and ways forward 31 Trust:Self-positioning rubric Table 5.Rubric for building trust between key players around generative AI.Emerging Established Evolved Extending Leaders Planning and initiating conversations on AI use and impacts.Preliminary engagement with AI vendors.Developing basic AI

265、governance.Clear principles,rules,and feedback mechanisms for AI use.Establishing basic data privacy and security measures.Regularly engage with educators on AI use.Some risk maturity to support AI experimentation.Establishing some oversight mechanisms.Fostering an environment that supports safe and

266、 responsible AI experimentation and learning.Collaborates with educators and students on AI use.Comprehensive AI vendor engagement processes.Engages with some partners on AI use.Formal oversight and evaluation mechanisms with clear accountability lines.Pioneering adaptive AI governance models.Influe

267、ncing peer institutions and/or national conversations between key stakeholder groups.Actively engages with partners(industry,professional bodies,community,alumni,government)on AI use expectations.Educators Cautious exploration of AI use cases.Lacks transparency around own use of AI.Seeking clarity o

268、n policies.Transparency about own use of AI.Openly discussing AI use with students and colleagues.Actively partnering with students and peers to develop AI literacy.Modelling and promoting transparent and ethical AI use.Co-creating AI rules,practices,and ecosystem with leaders,peers,and students.Bri

269、dging industry needs with curriculum.Researchers Cautious exploration of AI use cases.Lacks transparency around own use of AI.Seeking clarity on policies.Clear documentation of AIs role in research methods.Sharing of AI experiences with research peers.Actively contributing to institutional AI trust

270、guidelines.Mentoring and modelling of transparent and ethical use of AI in research practices.Pioneering methods for evaluating and validating AI use in research practices.Collaborating with industry and peers on AI trustworthiness in research.Students Initial guided use of AI applications.Tentative

271、 trust in institution-provided AI resources.Guarded about AI use.Engaging in discussions around responsible AI use.Trusting institution-provided AI applications.Transparency about own use of AI with peers and educators.Developing mindful trust in AI outputs.Critically evaluating AIs strengths and li

272、mitations,and impact on learning.Balancing AI assistance with personal skill development.Open encouragement of peers to use AI.Co-designing AI-enhanced learning experiences with educators.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 32

273、Culture The final,and arguably the most complex,element of CRAFT lies in culture.This is multi-faceted and includes(i)regional,geographical,and societal responses to technology and automation,(ii)institutional or departmental cultures around innovation,collaboration,and risk,(iii)disciplinary reacti

274、ons to generative AI,and(iv)a more wholesale consideration around the role of the university.First,regionally and geographically across the Pacific Rim,there are differences in perceptions of risk and benefit of AI systems.For example,a recent report76 suggests that people in China and Singapore app

275、ear to be most optimistic about AI and perceive that the benefits outweigh the risks,whereas people in the US,Canada,and Australia,and to an extent Japan and South Korea,are less positive this tends to follow the level of AI use at work and perceptions of employer support for AI.The report authors a

276、lso suggest that those from emerging economies may have a stronger cultural acceptance of technology as it may be perceived as a route towards economic progress and advancement.There are also cultural differences in how teacher authority is perceived between Western and Eastern educational philosoph

277、ies.It remains an open question whether AI may be seen to erode a traditional teacher-student dynamic in Confucian education cultures,or whether education cultures that promote more critical and independent thought and questioning of authority might respond differently to the effects of generative A

278、I.For example,in Western education systems that typically prioritize student autonomy and creativity,would the use of generative AI tend more towards exploratory applications?Or,would Confucian systems prioritize the application of generative AI applications where the teacher maintains more control

279、over AI,perhaps with AI deliberately designed to take the role of a Confucian teacher?These are areas that are worth exploring further when considering cultural intersections with generative AI.Secondly,across different institutions and departments there are variable appetites for risk,experimentati

280、on,and collaboration.As already stated,collegial exploration is needed to discover productive and responsible ways to use generative AI in context.To support this culture of experimentation,the right rules,access,and(to an extent)foundational familiarity need to be established,providing staff and st

281、udents with an 76 Gillespie et al.(2023)Trust in Artificial Intelligence:Global Insights 2023.https:/ 77 Singapore Government National AI Strategy.https:/www.smartnation.gov.sg/nais/Case study The Singapore Government released77 its National AI Strategy 2.0 in 2023,bringing together citizens,busines

282、ses,researchers,and the government to enhance national capability and infrastructure around AI.Since the first national AI strategy in 2019,significant investment has seen a rapid expansion of AI applications and enablement activity including research and start-ups.The new strategy focuses on buildi

283、ng familiarity(seeing AI as a“must know”),forming global alliances and partnerships to contribute to AI development,and scaling out AI-enabled solutions across the economy.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 33 environment for

284、safe exploration without fear of unfair reprisal78.Even though risk maturities vary widely between institutions,there are many common elements on which universities can and should collaborate.For example,the Higher Education Community Vendor Assessment Toolkit is a shared framework for institutions

285、to gauge vendor risk,since potential risk concerns are mostly common between institutions79.EDUCAUSE and other groups have established lively online communities where leaders,educators,and researchers can share resources,events,and experiences80.Similarly,risks and opportunities around AI use in edu

286、cation(such as in assessment)are also common,so the sharing of approaches and policies across institutions will help the sector avoid repeating missteps81.For example,the Australian Governments Tertiary Education Quality and Standards Agency has collaborated with assessment and AI experts to provide

287、 sector-wide guidance around assessment reform82,and Australian university learning and teaching leaders have had regular national roundtables to share practices around generative AI83.Students are another obvious collaborative partner especially regarding AI and education.Students as partners initi

288、atives were already increasing in prevalence across the sector in recent years;this shift in culture and build-up of momentum needs to be leveraged so that students as seen as equal partners in responding to AI.A key risk is that a prevailing culture of institutional competition and exceptionalism i

289、s likely to lead to reinventing the wheel many times over,such as already visible through multiple institutions across the regions building their own custom AI platforms and AI-driven avatar tools.More collaboration and partnerships across and within the higher education sector,and with community an

290、d industry,will benefit all institutions and their communities.Different disciplinary and industry cultures will also react differently to the capabilities of generative AI.Early reflections suggest that while there are commonalities between disciplines(such as considerations of efficiency gains and

291、 technical limitations),there may be industry-by-industry nuances that impact how 78 McDonald et al.(2024)Apostles,Agnostics and Atheists:Engagement with Generative AI by Australian University Staff.https:/eprints.qut.edu.au/252079/79 EDUCAUSE(2024)Higher Education Community Vendor Assessment Toolki

292、t.https:/library.educause.edu/resources/2020/4/higher-education-community-vendor-assessment-toolkit 80 For example,EDUCAUSEs AI Community Group(https:/connect.educause.edu/community-home/digestviewer?CommunityKey=3e9c1d98-f63e-4ac4-9efd-0187b8b72c8a)and the AI in Education Google group(https:/ Rober

293、t and McCormack(2024)2024 EDUCAUSE Action Plan:AI Policies and Guidelines.https:/www.educause.edu/research/2024/2024-educause-action-plan-ai-policies-and-guidelines 82 Lodge et al.(2023)Assessment reform for the age of artificial intelligence.https:/www.teqsa.gov.au/guides-resources/resources/corpor

294、ate-publications/assessment-reform-age-artificial-intelligence 83 Liu et al.(2023)Working paper:Responding to Generative AI in Australian Higher Education.https:/osf.io/preprints/edarxiv/9wa8p Case study Technolgico de Monterrey is shifting institutional culture by supporting a series of projects th

295、at leverage AI for teaching and learning,research and development,and operations.Taking a principles-based approach with values including respect for human dignity,equity,transparency,and autonomy,Tec is collaborating with researchers,educators,industry,and other organizations on applying AI for hea

296、lthcare,student success,personalized learning,systems navigation,and developing AI literacies in graduates.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 34(and how much)generative AI is accepted84.For example,financial and healthcare dis

297、ciplines may be more concerned with accuracy and liability,compared with management and business that may raise more issues around automation and worker displacement.There may also be differences between functions within organizations(such as marketing vs sales vs human resources)85.Within academia,

298、the textual nature of many generative AI outputs may be perceived as more of an affront to humanities disciplines,which may be reflected in how much AI is currently used by different disciplines(e.g.more in engineering and information technology compared to society and culture86).As we work to produ

299、ctively and responsibly engage with generative AI in higher education,we need to be compassionately mindful of the fundamental knowledge,skills,and dispositions that different disciplines hold dear and find hardest to concede.As with other aspects of culture,further investigation is needed to consid

300、er intersections of academic disciplinary culture with perspectives and reactions to generative AI.Finally,the generative AI conversation in higher education has shifted somewhat over the past two years from panic around academic integrity to a deeper reconsideration of purpose of higher education87

301、.The prevailing culture around the role of universities has been a perception that our institutions are bastions of knowledge creation and dissemination.However,generative AI has further democratized the access to knowledge,explanations,and interpretations that the internet had already accelerated.A

302、lthough renewed by generative AI,this conversational shift ventures beyond AI and into the polycrisis88 the sector is facing.Deeply held beliefs and concerns around the value of human expertise and the impersonal nature of AI-assisted learning are also powerful cultural factors to address89.Fundamen

303、tally,there needs to be a forward-looking culture that allows consideration of a future for universities that may look uncomfortably different from today in terms of the role of AI,the value placed on university credentials by employers,and traditional models of curriculum and the credit-hour90 that

304、 dictate the pace of advancement through programs.A key question that is increasingly being asked is:what is the role of universities,especially research-intensive universities that form the membership of APRU,and how should that evolve?Does the cultural mainstay of knowledge still hold,or do univer

305、sities need to refocus and rebalance towards what students can do,or who students become?In other words,and to alliterate,should universities focus on stuff(content,knowledge),skills(transferable skills),or soul(values,dispositions,beliefs,characteristics)(Figure 2)?Do we have the right balance of t

306、hese three elements in our 84 Dwivdei et al.(2023)Opinion Paper:“So what if ChatGPT wrote it?”Multidisciplinary perspectives on opportunities,challenges and implications of generative conversational AI for research,practice and policy.https:/doi.org/10.1016/j.ijinfomgt.2023.102642 85 IDC(2024)The Bu

307、siness Opportunity of AI report.https:/ 86 McDonald et al.(2024)Apostles,Agnostics and Atheists:Engagement with Generative AI by Australian University Staff.https:/eprints.qut.edu.au/252079/87 Joffres and Rey-Saturay(2024)The University at a Crossroads-Reimagining Higher Education in an Age of Disru

308、ption.https:/www.apru.org/resources_report/generative-ai-in-higher-education-foresight-workshop/88 World Economic Forum(2023)Were on the brink of a polycrisis how worried should we be?https:/www.weforum.org/stories/2023/01/polycrisis-global-risks-report-cost-of-living/89 Joffres and Rey-Saturay(2024

309、)Generative AI in Higher Education Sensemaking Workshop Proceedings.https:/www.apru.org/resources_report/generative-ai-in-higher-education-sensemaking-workshop/90 Joffres and Rey-Saturay(2024)The University at a Crossroads-Reimagining Higher Education in an Age of Disruption.https:/www.apru.org/reso

310、urces_report/generative-ai-in-higher-education-foresight-workshop/Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 35 courses and programs?What is the culture around what is valuable to gain from a higher education experience?Figure 2.Eleme

311、nts of stuff,skills,and soul when considering what students should be learning from their time at university.As we look to the future,especially as we question the role of universities,an important additional aspect of culture to consider is whether universities are preparing for powerful AI agentic

312、 AI that is as capable or more capable than human intelligence91-or perhaps even direct neural integration between mind and machine.Do we have a culture that looks far enough into the future so that we are preparing ourselves and our students for a radically transformed environment?Consider the foll

313、owing rubric(Table 6)to help situate your institutional and local progress and determine key action areas for development.91 Amodei(2024)Machines of Loving Grace.https:/ Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 36 Culture:Self-posit

314、ioning rubric Table 6.Rubric for fostering productive and responsible cultures around generative AI engagement.Emerging Established Evolved Extending Leaders Recognizing differing local/regional attitudes to technology.Acknowledging the digital divide in context.Identifying workforce AI needs.Aligni

315、ng AI strategy to local/regional educational philosophies.Implementing measures to address digital divides.Engaging with partners to understand AI skill needs.Identifying cultural misalignments between AI models and institutional contexts.Fostering an institutional culture of safe experimentation an

316、d failure.Sets the tone for institutional activities and aspirations.Supporting communities of practice and/or mentoring to support bottom-up culture change.Explicitly considering cultural elements in institutional strategies for AI.Pioneering culturally sensitive approaches to integrating AI.Leadin

317、g in ethical AI adoption across diverse cultural contexts.Fostering a future-looking culture to prepare for powerful AI92 including its implications for the purpose of university.Educators Exploring how AI fits within existing educational philosophies.Identifying discipline-specific challenges,barri

318、ers and stigma around AI.Adapting teaching methods to include AI while respecting cultural norms and expectations.Addressing discipline-specific concerns around AI use.AI use is destigmatized.Recognizing the cultural values embedded in AI models.Developing culturally appropriate AI pedagogies,and ad

319、vocating use amongst peers.Working with industry to align desired AI skills with curriculum.Responding to differing cultural values in AI models.AI use is widely accepted.Co-creating cross-institutional culturally sensitive AI education approaches.Pioneering new teaching approaches balancing AI and

320、core disciplinary values.Preparing for the implications of powerful AI on teaching and learning.Researchers Identifying field-specific barriers to AI adoption.Acknowledging cultural implications of AI applications in research practices.Adapting AI-enabled research practices to respect cultural norms

321、.Developing culturally sensitive protocols for AI use in research.Leading culturally informed AI-supported research practices.Fostering interactions between different research traditions and AI adoption.Shaping institutional or cross-institutional practices for culturally sensitive AI integration in

322、 research.Preparing for the implications of powerful AI on research.Students Becoming aware of local,disciplinary or cultural variations in AI perception,comfort and use.Engaging in culturally sensitive discussions on ethical AI use.Developing and embedding AI skills relevant to discipline.Encourage

323、d to demonstrate their uses of AI.Awareness that AI models are shaped by their cultural origins.Critically examining role of AI in their discipline and cultural context.Contributing to shaping institutional AI culture.Co-leading initiatives to bridge cultural gaps in AI literacy while being cultural

324、ly sensitive.Preparing for the implications of powerful AI on work and society.92 We use powerful AI to mean AI that can operate at or beyond human levels of capability across a broad range of domains.In many contexts this has been referred to as artificial general intelligence.Association of Pacifi

325、c Rim Universities Generative AI in higher education:Current practices and ways forward 37 The importance of all five areas The CRAFT framework has been designed with three intersecting core elements(rules,access,familiarity)surrounded by trust and culture as supportive structures.The five component

326、s are interconnected and interacting,for example:Access and familiarity without rules this may lead to unsafe use of AI(such as inadvertently providing confidential information to AI vendors),or secret hidden use of AI,or challenges around the trustworthiness of higher education awards due to assess

327、ment practices where validity is not appropriately considered.These degrade trust(e.g.between people and AI,and between the community and universities)and set back development of a productive culture around AI.Access and rules without familiarity this may lead to rigid and basic use of AI without be

328、ing able to explore its potential,and people may use AI without understanding its ethical challenges leading to uncritical engagement with its outputs or poor pedagogical practices with AI.Similarly,these can degrade trust and may lead to a culture that is unable to look sufficiently forwards.Rules

329、and familiarity without access this may lead to a widening of the digital divide and exacerbation of inequity where only well-off students,educators,and researchers are able to access AI applications powered by frontier models.This has implications for academic integrity where some students will be

330、able to use AI to achieve better outcomes and prevents the development of a collective culture around AI.A lack of trust depending on which trust pairs are degraded,this can slow productive and responsible adoption of generative AI that then impacts culture.For example,over-focusing on academic inte

331、grity and taking a policing mindset erodes trust between students and educators and the institution.This can damage the development of a forward-looking culture that accepts and works with AI.Not having the right culture this can degrade collaboration and context-sensitive engagement with AI,as well

332、 as impacting the ability of institutions to plan ahead.Over time,this can erode trust between people,and trust of AI,as well as reduce motivation to develop or maintain familiarity.All five elements of the CRAFT framework are necessary to enable individuals and institutions to move ahead with gener

333、ative AI.Whilst no framework is completely exhaustive,CRAFT encapsulates the essential elements needed to make practical progress.Association of Pacific Rim Universities Generative AI in higher education:Current practices and ways forward 38 Looking ahead The CRAFT model synthesizes a practical and scaffolded way for institutions and the sector to respond to generative AI responsibly,systematicall