法规注册部方面的AI应用

Taskforce announced in the Integrated Review Refresh2023130will support government to buildUK capability and ensure the UK harnesses the benefits presented by this emerging technology.The value chain of general-purpose AI,Ada Lovelace Institute,2023.See for example,The value chain of general-purpose AI,Ada Lovelace Institute,2023; An overview of AI alignment,Conjecture,2023; Make safe systems and deploy them reliably,Anthropic,2023.Integrated Review Refresh 2023,Prime Minister’s Office,10 Downing Street,Foreign,Commonwealth and DevelopmentOffice,Ministry of Defence 2023.A pro-innovation approach to AI regulationOur proposed framework will ensure we create the right regulatory environment as we move tomaximise the transformative potential of foundation models.Case-study 3.9:Life cycle accountability for large language modelsLarge language models(LLMs)are a type of foundationmodel.131The potential of LLMsgoes beyond reproducing or translating natural language:LLMs also have the power towritesoftware,132generatestories133through films and virtualreality,134andmore.135LLMs fall within the scope of our regulatory framework as they are autonomous andadaptable.We are mindful of the rapid technological change in the development of foundation modelssuch as LLMs and the new opportunities that they bring to applications including searchengines,medical devices,and financial and legal services.However,LLMs also havelimitations,for example,the models are not trained on a sense oftruth,136so they canreproduce inconsistent or false outputs that seem highly

As we find more uses for AI,it will rewrite scientificfields and change the way we learn about our world.Deep learning AI could improve breast cancer screeningAI could transform how diseases are detected,prevented,and treated.Doctors aretesting if deep learning can be applied to breast cancer screening.Currently,everymammogram is double-checked by radiologists but this is labour-intensive andcauses diagnosis delays.A UK medical technology company is working with theSee,for example,What are Large Language Models used for?NVIDIA,2023.Black hole pictured for first time – in spectacular detail,Nature,2019.Accelerating the race against antibiotic resistance,Deepmind,2022.Stopping malaria in its tracks,Deepmind,2022.Creating plastic-eating enzymes that could save us from pollution,Deepmind,2022.A pro-innovation approach to AI regulationNHS to test AI for the second screening,meaning greater numbers of patientscould be screened faster and clinicians could spend more time with patients andprovide faster access totreatment.33Farming efficiency increased by AI robotsApplying robotics and AI to field management can make farming more efficient,sustainable and productive.Lightweight,autonomous mapping and monitoringrobots operating across the UK can spend hours on the field in all conditions andsignificantly reduce soil compaction.These systems can digitise the field,providingfarmers with data to improve weed and pest management.If these systemsbecome widely used,they could contribute to agricultural and horticulturalproductivity,reduce the pressure of labour shortages and better preserve the

Scientists may also have succeeded in using generative AI to designantibodies that bind to a human protein linked tocancer.36AI is used in the fight against the most serious and harmful crimesThe Child Images AbuseDatabase37uses the powerful data processingcapabilities of AI to identify victims and perpetrators of child sexual abuse.Thequick and effective identification of victims and perpetrators in digital abuse imagesMia mammography intelligent assessment,NHS England,2021.Robotics and Autonomous Systems for Net Zero Agriculture,Pearson et al.,2022.Artificial intelligence,big data and machine learning approaches to precision medicine and drug discovery,Current DrugTargets,2021.Unlocking de novo antibody design with generative artificial intelligence,Shanehsazzadeh et al.,2023.Pioneering new tools to be rolled out in fight against child abusers,Home Office,2019.A pro-innovation approach to AI regulationallows for real world action to remove victims from harm and ensure their abusersare held to account.The use of AI increases the scale and speed of analysis whileprotecting staff welfare by reducing their exposure to distressing content.AI increases cyber security capabilitiesCompanies providing cyber security services are increasingly using AI to analyselarge amounts of data about malware and respond to vulnerabilities in networksecurity at faster-than-humanspeeds.38As the complexity of the cyber threatlandscape evolves,the pattern-recognition and recursive learning capabilities of AIare likely to play an increasingly significant role in proactive cyber defence againstmalicious actors.1.2 Managing AI risks