《OpenAI：2025 AI在企業中的應用研究報告（英文版）（25頁）.pdf》由會員分享，可在線閱讀，更多相關《OpenAI：2025 AI在企業中的應用研究報告（英文版）（25頁）.pdf（25頁珍藏版）》請在三個皮匠報告上搜索。

1、AI in the EnterpriseLessons from seven frontier companiesContentsA new way to work3Executive summary5Seven lessons for enterprise AI adoptionStart with evals6Embed AI into your products9Start now and invest early11Customize and fine-tune your models13Get AI in the hands of experts16Unblock your deve

2、lopers18Set bold automation goals21Conclusion22More resources242AI in the EnterpriseA new way to workAs an AI research and deployment company,OpenAI prioritizes partnering with global companies because our models will increasingly do their best work with sophisticated,complex,interconnected workflow

3、s and systems.Were seeing AI deliver significant,measurable improvements on three fronts:01Workforce performanceHelping people deliver higher-quality outputs in shorter time frames.02Automating routine operationsFreeing people from repetitive tasks so they can focus on adding value.03Powering produc

4、tsBy delivering more relevant and responsive customer experiences.3AI in the EnterpriseBut leveraging AI isnt the same as building software or deploying cloud apps.The most successful companies are often those who treat it as a new paradigm.This leads to an experimental mindset and an iterative appr

5、oach that gets to value faster and with greater buy-in from users and stakeholders.Our approach:iterative developmentOpenAI is organized around three teams.Our Research Team advances the foundations of AI,developing new models and capabilities.Our Applied Team turns those models into products,like C

6、hatGPT Enterprise and our API.And our Deployment Team takes these products into companies to address their most pressing use cases.We use iterative deployment to learn quickly from customer use cases and use that information to accelerate product improvements.That means shipping updates regularly,ge

7、tting feedback,and improving performance and safety at every step.The result:users access new advancements in AI early and oftenand your feedback shapes future products and models.4AI in the EnterpriseExecutive summarySeven lessons for enterprise AI adoption01Start with evalsUse a systematic evaluat

8、ion process to measure how models perform against your use cases.02Embed AI in your productsCreate new customer experiences and more relevant interactions.03Start now and invest earlyThe sooner you get going,the more the value compounds.04Customize and tune your modelsTuning AI to the specifics of y

9、our use cases can dramatically increase value.05Get AI in the hands of expertsThe people closest to a process are best-placed to improve it with AI.06Unblock yourdevelopers Automating the software development lifecycle can multiply AI dividends.07Set bold automation goals Most processes involve a lo

10、t of rote work,ripe for automation.Aim high.Lets drill down into each of these,with customer stories as examples.5AI in the EnterpriseLesson 1Start with evalsHow Morgan Stanley iterated to ensure quality and safetyAs a global leader in financial services,Morgan Stanley is a relationship business.Not

11、 surprisingly,there were some questions across the business about how AI could add value to the highly personal and sensitive nature of the work.The answer was to conduct intensive evals for every proposed application.An eval is simply a rigorous,structured process for measuring how AI models actual

12、ly perform against benchmarks in a given use case.Its also a way to continuously improve the AI-enabled processes,with expert feedback at every step.How it startedMorgan Stanleys first eval focused on making their financial advisors more efficient and effective.The premise was simple:If advisors cou

13、ld access information faster and reduce the time spent on repetitive tasks,they could offer more and better insights to clients.They started with three model evals:01Language translationMeasuring the accuracy and quality of translations produced by a model.02SummarizationEvaluating how a model conde

14、nses information,using agreed-upon-metrics for accuracy,relevance,and coherence.03Human trainersComparing AI results to responses from expert advisors,grading for accuracy and relevance.These evalsand othersgave Morgan Stanley the confidence to start rolling the use cases into production.6AI in the

15、EnterpriseHow its goingToday,98%of Morgan Stanley advisors use OpenAI every day;access to documents has jumped from 20%to 80%,with dramatically reduced search time;and advisors spend more time on client relationships,thanks to task automation and faster insights.The feedback from advisors has been o

16、verwhelmingly positive.Theyre more engaged with clients,and follow-ups that used to take days now happen within hours.Kaitlin ElliottHead of Firmwide Generative AI SolutionsTo find out more,watch Morgan Stanley:Shaping the Future of Financial Services and ask us about our Eval Frameworks.7AI in the

17、EnterpriseEvals definedEvaluation is the process of validating and testing the outputs that your models produce.Rigorous evals lead to more stable,reliable applications that are resilient to change.Evals are built around tasks that measure the quality of the output of a model against a benchmarkis i

18、t more accurate?More compliant?Safer?Your key metrics will depend on what matters most for each use case.8AI in the EnterpriseLesson 2Embed AI into your products How Indeed humanizes job matchingWhen AI is used to automate and accelerate tedious,repetitive work,employees can focus on the things only

19、 people can do.And because AI can process huge amounts of data from many sources,it can create customer experiences that feel more human because theyre more relevant and personalized.Indeed,the worlds No.1 job site,uses GPT-4o mini to match job seekers to jobs in new ways.The power of whyMaking grea

20、t job recommendations to job seekers is only the start of the Indeed experience.They also need to explain to the candidate why this specific job was recommended to them.Indeed uses the data analysis and natural language capabilities of GPT-4o mini to shape these why statements in their emails and me

21、ssages to jobseekers.Using AI,the popular Invite to Apply feature also explains why a candidates background or previous work experience makes the job a good fit.The Indeed team tested the previous job matching engine against the GPT-powered version with the new,customized context.The performance upl

22、ift was significant:A 20%increase in job applications startedA 13%uplift in downstream successnot only were more candidates likely to apply,but employers were more likely to hire them.9AI in the EnterpriseWith Indeed sending over 20 million messages a month to job seekersand 350 million visitors com

23、ing to the site every monththese increases scale up to significant business impact.But scaling up also meant using more tokens.To increase efficiency,OpenAI and Indeed worked together to fine-tune a smaller GPT model that was able to deliver similar results with 60%fewer tokens.Helping job seekers f

24、ind the right jobsand understanding why a given opportunity is right for themis a profoundly human outcome.Indeeds team has used AI to help connect more people to jobs,fastera win for everyone.We see a lot of opportunity to continue to invest in this new infrastructure in ways that will help us grow

25、 revenue.Chris HyamsCEO10AI in the EnterpriseLesson 3Start now and invest earlyHow Klarna benefits from AI knowledge compoundingAI is rarely a plug-and-play solutionuse cases grow in sophistication and impact through iteration.The earlier you start,the more your organization benefits from compoundin

26、g improvements.Klarna,a global payments network and shopping platform,introduced a new AI assistant to streamline customer service.Within a few months,the assistant was handling two-thirds of all service chatsdoing the work of hundreds of agents and cutting average resolution times from 11 minutes t

27、o just 2.The initiative is projected to deliver$40 million in profit improvement,all while maintaining satisfaction scores on par with human support.These results didnt happen overnight.Klarna achieved this performance by continuously testing and refining the assistant.Just as importantly,90%of Klar

28、nas employees now use AI in their daily work.Growing organization-wide familiarity with AI has enabled Klarna to move faster,launch internal initiatives more efficiently,and continuously refine the customer experience.By investing early and encouraging broad adoption,Klarna is seeing AIs benefits co

29、mpounddriving returnsacross its business.11AI in the EnterpriseThis AI breakthrough in customer interaction means superior experiences for our customers at better prices,more interesting challenges for our employees,and better returns for our investors.Sebastian SiemiatkowskiCo-Founder and CEO12AI i

30、n the EnterpriseLesson 4Customize and fine-tune your modelsHow Lowes improves product searchEnterprises seeing the most success from AI adoption are often the ones that invest time and resources in customizing and training their own AI models.OpenAI has invested heavily in our API to make it easier

31、to customize and fine-tune modelswhether as a self-service approach or using our tools and support.We worked closely with Lowes,the Fortune 50 home improvement company,to improve the accuracy and relevance of their ecommerce search function.With thousands of suppliers,Lowes often has to work with in

32、complete or inconsistent product data.13AI in the EnterpriseThe key is in accurate product descriptions and tagging.But it also requires an understanding of how shoppers search,a dynamic that changes across product categories.Thats where fine-tuning comes in.By fine-tuning OpenAI models,the Lowes te

33、am was able to improve product tagging accuracy by 20%with error detection improving by 60%.Excitement in the team was palpable when we saw results from fine-tuning GPT 3.5 on our product data.We knew we had a winner on our hands!Nishant GuptaSenior Director,Data,Analytics and Computational Intellig

34、enceProduct Note:OpenAI has launched Vision Fine-Tuning to further improve ecommerce search and address challenges in medical imaging and autonomous driving.14AI in the EnterpriseWhat is fine-tuning?If a GPT model is a store-bought suit,fine-tuning is the tailored optionthe way you customize the mod

35、el to your organizations specific data and needs.Why it matters:Improved accuracyBy training on your unique datasuch as product catalogs or internal FAQsthe model delivers more relevant,on-brand results.Domain expertiseFine-tuned models better understand your industrys terminology,style,and context.

36、Consistent tone and styleFor a retailer,that could mean every product description stays true to brand voice;for a law firm,it means properly formatted citations,every time.Faster outcomesLess manual editing or re-checking means your teams can focus on high-value tasks.15AI in the EnterpriseLesson 5G

37、et AI in the hands of expertsBBVA takes an expert-led approach to AIYour employees are closest to your processes and problems and are often the best-placed to find AI-driven solutions.Getting AI into the hands of these experts can be far more powerful than trying to build generic or horizontal solut

38、ions.BBVA,the global banking leader,has more than 125,000 employees,each with a unique set of challenges and opportunities.They decided to get AI into the hands of employeesworking closely with Legal,Compliance,and IT Security teams to ensure responsible use.They rolled out ChatGPT Enterprise global

39、ly,then let people discover their own use cases.“Normally,in a business like ours,building even a prototype requires technical resources and time,”says Elena Alfaro,Head of Global AI Adoption at BBVA.“With custom GPTs,anyone can create apps to solve unique problemsits very easy to start.”In five mon

40、ths,BBVA employees created over 2,900 custom GPTssome of which reduce project and process timelines from weeks to hours.The impact was felt across many disciplines and departments:The Credit Risk teamUses ChatGPT to determine creditworthiness faster and more accurately.The Legal teamUses it to answe

41、r 40,000 questions a year on policies,compliance,and more.The Customer Service teamAutomates the sentiment analysis of NPS surveys.16AI in the EnterpriseAnd the wins continue to spread across Marketing,Risk Management,Operations,and beyond.All because they got AI in the hands of the people who know

42、how to apply it in their own disciplines.We consider our investment in ChatGPT an investment in our people.AI amplifies our potential and helps us be more efficient and creative.Elena AlfaroHead of Global AI AdoptionProduct Note:With deep research,ChatGPT can do work independently.Give it a prompt,a

43、nd it can synthesize hundreds of online sources to create comprehensive,PhD-level reports.This unlocks employee productivity and gives them access to deep,detailed research on any topic in minutes.In an internal evaluation by experts across domains,deep research saved an average of 4 hours per compl

44、ex task.For more detail,watch BBVA puts AI into the hands of every team.17AI in the EnterpriseLesson 6Unblock your developersMercado Libre builds AI programs faster and more consistentlyDeveloper resources are the main bottleneck and growth inhibitor in many organizations.When engineering teams are

45、overwhelmed,it slows innovation and creates an insurmountable backlog of apps and ideas.Mercado Libre,Latin Americas largest ecommerce and fintech company,partnered with OpenAI to build a development platform layer to solve that.Its called Verdi,and its powered by GPT-4o and GPT-4o mini.Today,it hel

46、ps their 17,000 developers unify and accelerate their AI application builds.Verdi integrates language models,Python nodes,and APIs to create a scalable,consistent platform that uses natural language as a central interface.Developers now build consistently high-quality apps,faster,without having to g

47、et into the source code.Security,guardrails,and routing logic are all built in.18AI in the EnterpriseAs a result,AI app development has accelerated dramatically,helping Mercado Libre employees do amazing things,including:Improving inventory capacityGPT-4o mini Vision tags and completes product listi

48、ngs,allowing Mercado to catalog 100 x more products.Detecting fraudEvaluating data on millions of product listings each day,improving fraud detection accuracy to nearly 99%for flagged items.Customizing product descriptionsTranslating product titles and descriptions to adapt to nuanced Spanish and Po

49、rtuguese dialects.Increasing ordersAutomating review summaries to help users quickly grasp product feedback.Personalizing notificationsTailoring push notifications to drive higher engagement and improve product recommendations.19AI in the EnterpriseNext up:using Verdi to improve logistics,reduce lat

50、e deliveries,and take on more high-impact tasks across the organization.We designed our ideal AI platform using GPT-4o mini,with a focus on lowering cognitive load and enabling the entire organization to iterate,develop,and deploy new,innovative solutions.Sebastian BarriosSVP of Technology20AI in th

51、e EnterpriseLesson 7Set bold automation goalsHow we automate our own work at OpenAIAt OpenAI,we live with AI every day,so were often spotting new ways to automate our own work.An example:Our support teams were getting bogged down,spending time accessing systems,trying to understand context,craft res

52、ponses,and take the right actions for customers.So we built an internal automation platform.It works on top of our existing workflows and systems to automate rote work and accelerate insight and action.Our first use case:working on top of Gmail to craft customer responses and trigger actions.Using o

53、ur automation platform,our teams can instantly access customer data and relevant knowledge articles,then incorporate the results into response emails or specific actionssuch as updating accounts or opening support tickets.By embedding AI into existing workflows,our teams are more efficient,responsiv

54、e,and customer-focused.This platform handles hundreds of thousands of tasks every month,freeing people to do more high-impact work.Not surprisingly,the system is now spreading across other departments.It happened because we set bold automation goals from the start,instead of accepting inefficient pr

55、ocesses as a cost of doing business.21AI in the EnterpriseConclusionLearning from each otherAs the previous examples show,every business is full of opportunities to harness the power of AI for improved outcomes.The use cases may vary by company and industry but the lessons apply across all markets.T

56、he common theme:AI deployment benefits from an open,experimental mindset,backed by rigorous evaluations,and safety guardrails.The companies seeing success arent rushing to inject AI models into every workflow.Theyre aligning around high-return,low-effort use cases,learning as they iterate,then takin

57、g that learning into new areas.The results are clear and measurable:faster,more accurate processes;more personalized customer experiences;and more rewarding work,as employees focus on the things people do best.Were now seeing companies integrating AI workflows to automate increasingly sophisticated

58、processesoften using tools,resources,and other agents to get things done.Well continue to report back from the front lines of AI to help guide your own thinking.Product Note:Operator Operator is an example of OpenAIs agentic approach.Leveraging its own virtual browser,Operator can navigate the web,c

59、lick on buttons,fill in forms,and gather data just like a human would.It can also run processes across a wide range of tools and systemsno need for custom integrations or APIs.Enterprises use it to automate workflows that previously required human intervention,such as:Automating software testing and

60、 QA using Operator to interact with web apps like a real user,flagging any UI issues.Updating systems of record on behalf of users,without technical instructions or API connections.The result:end-to-end automation,freeing teams from repetitive tasks and boosting efficiency across the enterprise.22AI

61、 in the EnterpriseThe trusted AI enterprise platformSecurity and privacy at a glanceFor our enterprise customers,nothing is more important than security,privacy and control.Heres how we ensure it:Your data stays yoursWe dont use your content to train our models;your enterprise retains full ownership

62、.Enterprise-grade complianceData is encrypted in transit and at rest,aligned with top standards like SOC 2 Type 2 and CSA STAR Level 1.Granular access controlsYou choose who can see and manage data,ensuring internal governance and compliance.Flexible retentionAdjust settings for logging and storage

63、to match your organizations policies.For more on OpenAI and security,visit our Security page or the OpenAI Security Portal.23AI in the EnterpriseMore resourcesOpenAI for BusinessOpenAI StoriesChatGPT EnterpriseOpenAI and SafetyAPI PlatformOpenAI is an AI research and deployment company.Our mission is to ensure that artificial general intelligence benefits all of humanity.24AI in the Enterprise