AI驱动的研究、开发与创新

EUREKA!ON STEROIDS AI-driven research, development,and innovation “AI willacceleratescientificresearchfarmore than wecan imagine.” — Joëlle Barral, Director of FundamentalResearch in AI, Google DeepMind EUREKA! ON STEROIDS AI-driven research, development, and innovation Authors Dr. Albert Meige,Director of Blue Shift, Arthur D. LittleZoe Huczok,Project Leader of Blue Shift, Arthur D. LittleArnaud Siraudin,Associate Director, Arthur D. LittleDonatello Fleck,Business Analyst, Arthur D. LittleGeoffroy Barruel,Consultant, Arthur D. Little Contributors Pierre Blouet,Research, Development, andInnovation Director, GRTgazCarole Caranta,Deputy Director General of Scienceand Innovation, INRAEPaul-Joël Derian,Group VP Innovation andSustainable Development, AvrilPhilippe Mauguin,CEO, INRAEJean-Luc Moullet,Chief of Staff, French Minister ofHigher Education and ResearchChristophe Perthuisot,Chief Research & Innovation Officer,Moët HennessyAntoine Petit,President & CEO, CNRSRick Eagar,Partner Emeritus, Arthur D. Little Expert-in-residence Anne Bouverot,French President’s Special Envoy on AI Executive summary6Preamble12Chapter 1. The potential of AI in R&D&I16Chapter 2. How to ensure success24Interlude: Focus on data!36Chapter 3. Tools & providers38Chapter 4. Navigating the future44Chapter 5. Strategic actions64 Executivesummary Although AI has been used in specific research, development, andinnovation (R&D&I) applications for at least a decade, it’s been twoyears since the recent acceleration began, initiated by the availabilityof more powerful generative AI (GenAI) and large language models(LLMs). While there is a glut of information on potential applications,widespread integration of AI into R&D&I processes is still relativelyimmature. Applying AI to many R&D&I use cases poses significantchallenges, especially where outcomes need to be error-free, aswell as uncertainties in how AI will evolve regarding technology,economics, regulation, and societal acceptability. This in-depth study was conducted by Arthur D. Little’s (ADL’s) BlueShift in partnership with five major leading public and private sectororganizations already using AI in their R&D&I efforts: LVMH, Avril,GRTgaz, the French National Centre for Scientific Research (CNRS),and the French National Research Institute for Agriculture, Food,and the Environment (INRAE). The study explored the current stateof AI in R&D&I, the challenges and best practices, the landscapeof solution providers, and future scenarios. We gathered evidencethrough 40+ interviews from AI providers, independent AI experts,and current best-in-class users of AI in R&D&I, as well as a surveywith over 200 responses from private companies and publicinstitutions that examined AI maturity, contributions, benefits,andbarriers. HOW TO ENSURE SUCCESS THE POTENTIAL OF AI IN R&D&I Chapter 1: AI augments researchers’ capabilitiesacross all steps of the R&D&I process throughvarious roles, helping to solve intractable problemsand make decisions. No blanket model exists; dataavailability and problem type determine the bestmethod. Most often, AI models are embedded in asystems of systems.- Chapter 2: Ensuring success in AI implementationfor R&D&I requires agile methodologies, robust datafoundations, strategic prioritization, analyticaltradeoffs, scarce data science talent management,IT alignment, rapid benefit demonstration, andcontinuous monitoring.- Agile methodologies.Agile methodologies thatmove fast and iteratively are preferable for AI projectdevelopment, given the speed at which technologyevolves. Such approaches ensure that some benefitscan be obtained early, even if “perfect” solutions arestill some way off.- Benefits abound for AI.AI augments researchers’capabilities, rather than replacing them, as partof a people-centric R&D&I effort. It helps solveintractable problems that researchers couldn’ttackle before. It already acts as a knowledgemanager, hypothesis generator, and assistant. Theplanner/thinker archetype, in which AI helps makedecisions, is rapidly emerging.- Robust data foundations.Data collection hygiene,storage, security, and governance are central torealizing AI benefits. New techniques for processingpoorly structured or smaller datasets are becomingmore important. Ensuring wide data accessibility,cross-organizational collaboration, and effectivedata governance is also fundamental.- AI-based models support use cases at every stepof R&D&I process.These range from technology andmarket intelligence to innovation strategy, ideation,portfolio and project management, IP management,ecosystem management, knowledge management,and new product/service launch and deployment.- Strategic prioritization.Organizations mustchoose strategically between making, buying off-the-shelf, or fine-tuning AI models. Most core R&D&Iproblems lend themselves to fine-tuning existingopen source models, whether LLMs, generativeadversarial networks (GANs), diffusion models, orreinforcement learning (