《仲量聯行：2024商業地產行業AI投資研究報告：宏觀經濟下的AI戰略分析（英文版）（15頁）.pdf》由會員分享，可在線閱讀，更多相關《仲量聯行：2024商業地產行業AI投資研究報告：宏觀經濟下的AI戰略分析（英文版）（15頁）.pdf（15頁珍藏版）》請在三個皮匠報告上搜索。

1、Breaking ground:How CRE investors are embracing AI for real results In the current economic macroclimate,CRE investors should consider AI as a feasible way forwardRemain resilient in shaky market conditionsInterest rates are remaining stubbornly high,putting downward pressure on property values.Infl

2、ation is persistent and eating away at net operating income health.And most critically for us,the hybrid work revolution has made attendance levels of buildings chronically low.For instance,across all global regions,less than 45%of organizations surveyed by JLL reported employees coming into the off

3、ice five days a week.This is shrinking the demand for space,particularly offices,in a way that challenges the foundation of investor business models.The old real estate investment model that relied on low interest,middling occupancy levels,and highly consistent attendance is no longer feasible.1Cont

4、rol costs This can primarily be achieved by gleaning operational and strategic insights about where money is spent.Budgets are tight,and due to market conditions,risks are high.Rather than simply being seen as an experimental expenditure,AI should be seen as a means to alleviate the pain.For example

5、,Landsec is partnering with Microsoft to customize the GPT model and will target process efficiency gains across investment modelling,development,placemaking,leasing,and asset operations.2Boost staff productivity Fundamentally,AI solutions have been designed,and are being iterated,to be productivity

6、 tools.Deployed appropriately,AI tools enable users to interact more effectively with databases,informational documents,and collaborative platforms by quickly summarizing large sums of data.For example,more than 1,000 employees across Capital Markets and LaSalleboth divisions within JLL that deal wi

7、th investor clientsalready use the in-house JLL GPT tool on a weekly basis to complete their work tasks.They use it to automate email creation and accounting invoices as well as boost the efficiency of coding production.These are small tasks,but replicated across thousands of employees,this is a lar

8、ge-scale test case for boosting organizational productivity.3As the hype around generative AI turns into a more sober assessment of its true capabilities,stakeholders in the real estate market are starting to think about where it should be used first.For commercial real estate(CRE)investors in parti

9、cular,generative AI(and even less new AI practices)represent an opportunity to innovate.But why spare the expense?What are the key justifications?In short,there are three big reasons right now.CRE investors should investigate AI to:Four ways AI can demonstrably help:task productivity,outsourcing str

10、ategies,design automation,and data collectionThe interviewees for this research consistently pointed to the following use cases as the robust value adds.The analysis has been supplemented by JLL examples from its internal deployment of AI and product references from its partner network.These use cas

11、es enable near-term and significant returns as they focus on automating highly manual processes.Research methodology This report adopts a rigorous research approach,combining primary and secondary methodologies.The research framework involved leveraging JLLs internal research,reputable third-party s

12、ources,and internal consultants to develop a targeted survey.The aim was to gain valuable insights into industry stakeholders perceptions of AI solutions as a viable source of innovation.To identify the direction of market trends,we conducted interviews with a diverse group of 15 professionals,inclu

13、ding real estate finance analysts,asset development managers,and lecturers from public institutions.The information gathered from these interviews,along with further examination of secondary research,enabled us to confidently identify the key opportunities within the CRE investor market.One use case

14、 is the document review process.For example,a project manager at a real estate investment firm works with software partners who look at the design efficiency of new builds using generative AI.This tool helps identify potential discrepancies between spatial requirements and architectural concept draw

15、ings.If a specification outlines certain spatial dimensions for a lift,and a drawing shows divergence from this plan in its measurement,the software flags it.This solution can also highlight potential contradictions related to timeline commitments within a single,but long,planning document.Doing so

16、directly contributes to financial risk management objectives,helping prevent costly design mistakes through the supply chain that is needed to create the asset.Another use case is training a generative AI model on an internal database that monitors specific properties and suggests process adjustment

17、s.For example,Landsec says they have already deployed analytical AI methods in some of their back of house processes,reducing process time by 75%.One way they are doing this is by improving their building management systems(BMS)operations and trialling predictive as well as AI-driven technology to o

18、ptimize heating,ventilation,and air conditioning systems.This can also support ESG reporting on the portfolio.01Automate small parts of property management processes for substantive productivity gainsFrom the interviews,we heard that end users thought AI could help automate the delivery of outsource

19、d services,such as lease management,brokerage,and legal assistance during lease purchases.This cut across multiple aspects of administrative tasks,such as:Real estate and financial document searches:One area ripe for reworking is the abstraction,summarization,and tracking of changes to various docum

20、ent types.Since documents in real estate transactions are often very large,this can be a huge time saver.For instance,leasing teams at JLL are using APIs to extract numeric values from different clauses in their client invoices.Collectively,this is saving hours of time for these employees,which can

21、instead be spent on more complex tasks.Broker report delivery:AI can help speed up the turnaround time for brokerage reports.If you know that parts of the report creation process can be automated,the broker can skip to a review stage,speeding up the delivery to the end client.An AI model can also be

22、 trained to rapidly scan internal and online databases to spot properties of interest to a client,if it has detailed historical knowledge of previous sites they were interested in or purchased/leased.Terms and conditions of legal counsel services:AI can help reduce the time spent reviewing legal doc

23、uments.In turn,this can potentially reduce the billable hours you pay for the services.For instance,Orbital Witness(invested in by JLL Spark)can automate part of the legal document review process,thereby freeing up time for lawyers and other legal staff to focus on more complex tasks.The above are j

24、ust some examples of where AI can challenge existing outsourcing models.When AI can simplify a previously manual process and save time on delivery,customers should renegotiate terms and conditions to ensure they do not overpay.Their payment for services should instead be concentrated on the expertis

25、e-driven value that a generic AI model cannot replicate.02Re-examine the delivery process behind outsourced and insourced servicesA longstanding issue with deploying any type of monitoring equipment is the sheer volume of data it producesand the cost associated with properly interpreting it and then

26、 communicating insights to relevant stakeholders.The data from IoT sensors is often highly unstructured and needs a lot of processing before an end user can make sense of any patterns within it.During our interview fieldwork for this research,we learned that a UK-based land and asset development fir

27、m is using IoT sensors to measure the flow of water in pipes to identify potential sources of leakage.This is being achieved by using an analytical AI model and acts as a preventative measure to help future-proof assets,thereby improving net operating income and the financial health of development p

28、rojects.This is a good example of using AI technology to safeguard the structural integrity of a building.Additionally,Building Engines,a JLL company,recently announced that its Prism software solution can integrate with insurance risk solution Jones Software to optimize certificates of insurance,mi

29、nimizing the need for landlords,tenants,and vendors to send countless emails in their efforts to verify compliance.Prism uses Jones Softwares AI-powered two-phase verification,which identifies gaps in insurance documents and auto-generates emails to flag them.03Streamline both data and language revi

30、ews in asset risk monitoring activitiesHistorically,the biggest challenge with interrogating various pieces of information about a potential investment opportunity was organizing them in a way that enables fruitful interpretation.This would involve extensive data cleaning,formatting,and integrations

31、.Now,AI speeds up this process and simplifies the interface for interacting with the data.For example,Orbital Witness helps teams in the legal and property sectors store all the data related to due diligence for a property acquisition in one place.The solution enables them to locate and order inform

32、ation in a highly structured manner.Furthermore,by using generative AI to train its data model to recognize keywords and notions in the documents it reviews,it can search them more efficiently than a human.It can also use the insights to generate a range of report templates for communicating the inf

33、ormation to relevant stakeholders.As one interviewee put it,they would like to use a generative AI model to ask highly specific questions that would typically take multiple clicks and user-driven manipulation of multiple databases to identify.For instance,if a local investor in Edinburgh wants to id

34、entify the locations and characteristics of“townhouses nearby,”AI could provide a full report on this in a more structured,rapid manner than a user-driven search.04Train centralized data models to interrogate data more smartly8Three principles for approaching an AI deployment in CRE From our wider r

35、esearch,and from the interviews conducted for this report,the most high-impact opportunities are those that are highly defined.To define them well,you must situate the use of AI within your broader business capabilities,such as workflows that deliver front-or back-office processes.Often,it is better

36、 to start small,such as automating the creation of report templates in forecasting and due diligence processes,instead of the actual report content.If you can provably speed up an existing process and replicate it across multiple employees,this is an example of a high-impact deployment.At JLL,9%of p

37、roperty asset management staff were using JLL GPT,the firms in-house tool,at the end of December 2023,and by April 2024,it grew to 14%,showing the effectiveness of deployment efforts to grow adoption.More grandiose visions about using AI to automate the entire valuation process are,by contrast,likel

38、y to end in failure and minimize the opportunity to use AI where it is most appropriate right now.01Identify opportunities that are most likely to have a high impactOne of the key obstacles to a successful approach to AI is not understanding the risks associated with the task.These risks come in thr

39、ee categories:privacy,regulatory,and operational.The best approach to this is to set up governance policies for usage and testing out new capabilities.Importantly,the interview panel also saw data formatting and leaking as a key issue,too(see Figure 1 below).See our recent research paper for more de

40、tail on the risks associated with AI as well as how to mitigate them.02Cultivate an awareness of the risks when assessing market solutionsCost of deployment(licensing,integration,maintenance)73%Regulatory requirements64%Staff training costs53%Identifying and formatting data to analyze65%Risks relate

41、d to data leakage of internal insights55%Unclear business use cases39%Figure 1:Interviewee assessment of challenges related to AI deployment“How would you rank the following challenges related to using AI for investment activities?(Select three options only.)”It is important not to underestimate the

42、 unknowns in any innovation process.Training and fine-tuning analytical or generative AI on an internal database requires multiple rounds of testing.Use internal or outsourced experts to validate that the outputs are reliable and the reasoning behind an AIs analysis is consistently defendable.Since

43、this is likely to constitute the first step in an evaluation,as just one example,it must be credible before it can be adopted at scale.Therefore,expect multiple versions of an AI model to be incorrect and prepare to adjust its input regularly to continuously improve its outputs.As one interviewee pu

44、t it,“just as surveyors have a duty of care to ensure they make verifiable claims,any use of AI must follow the same principle;user input must be from reliable sources,and the outputs must mimic the integrity of a human analyst.”This is a critical element to justify the cost of deployment,with its m

45、ultiple components.Interviewees placed this as the biggest challenge.(see figure 1 on previous page).“Just as surveyors have a duty of care to ensure they make verifiable claims,any use of AI must follow the same principle;user input must be from reliable sources,and the outputs must mimic the integ

46、rity of a human analyst.Academic with real estate and technology expertise03Prepare to be agile during the execution journeyAppendixThis report breaks down how AI contributes to real estate objectives.This report is targeted at those involved in the CRE investment lifecycle,from land development and

47、 acquisition to asset management operations.They can be those in institutions and commercial enterprise,but also the self-employed and those with smaller businesses.In the analysis,we break down how deploying AI could feasibly contribute to the realization of the highly typical objectives that inves

48、tment firms orient their businesses around.These break down into:Financial risk managementPortfolio diversificationExit strategiesMaximizing net operating income(NOI)ESG reporting/obsolescence managementTo supplement the analysis made in this report,we interviewed more than 10 individuals in a range

49、 of positions,from analysts to developer managers and lecturers,across commercial and public institutions.The term AI is broad and can encompass both different techniques and practices that deploy those techniques.It is therefore necessary to outline precisely what we are referring to.In this report

50、,we cover two types of AI solutions.See Figure 2 below for a comparison of the two.What do we mean by AI?DefinitionAnalytical AIGenerative AIDeploys the AI technique of machine learning algorithms that review numeric-based data.The algorithms then continuously review new data and update their patter

51、n-recognizing efforts to keep the insights fresh.Either structured by humans or directly from a piece of equipment in an unstructured form to identify patterns.This is often done using probabilistic reasoning to reveal previously unseen patterns of correlation between data points.Dashboards displayi

52、ng historical trends,forecasted projections of trends,and various graphics summarizing highly categorized information.Uses large language models to interpret the meaning of language and numbers in a context of a specific request.These models are designed using neural networks that mimic the ability

53、of the human brain to process many layers of information.Relies on data to run through natural language processing(NLP).Various formats,including written text,pasted tables,and equations,can be ingested and referenced to answer user queries.Requires speech-processing technologies and computational l

54、inguistic techniques to interpret data.Written responses with accompanying reasoning.It can also reference external sources if the tool reviews information in the public realm.Data ingestion processOutcomesAuthorsIbrahim Yate Senior Research AnalystJLL TAkshay ThakurHead of Research&InsightsJLL TAbo

55、ut JLLFor over 200 years,JLL(NYSE:JLL),a leading global commercial real estate and investment management company,has helped clients buy,build,occupy,manage and invest in a variety of commercial,industrial,hotel,residential and retail properties.A Fortune 500company with annual revenue of$20.8 billio

56、n and operations in over 80 countries around the world,our more than 108,000 employees bring the power of a global platform combined with local expertise.Driven by our purpose to shape the future of real estate for a better world,we help our clients,people and communities SEE A BRIGHTER WAYSM.JLL is the brand name,and a registered trademark,of Jones Lang LaSalle Incorporated.For further information,visit .