谷歌开启无处不在的人工智能时代.pdf

1、Nathan PoelmanAI SpecialistGoogle CloudThe Everywhere AI Erawith GoogleAt Google,weve been working on AI for over a decadeNeural networklearns to recognisecats on YouTube2012Transformer2017AI Principles2016AlphaGo2015TensorFlow2024JasperClaudeChatGPTGeminiMistralMidjourneyLLaMaMission statementBuild

2、 AI responsibly to benefit humanity5Proprietary+Confidentialdesigned to be natively multimodalAI is not always called AI Inclusion through AIAssist especially in complex data domainsOur latest advances in robot dexterityGoogle AI PrinciplesAI ShouldAreas we will not pursue1likely to cause overall ha

3、rm2 principal purpose to direct injury3 surveillance violating internationally accepted norms4 purpose contravenes international law and human rights1be socially beneficial2avoid creating or reinforcing unfair bias3be built and tested for safety4be accountable to people5incorporate privacy design pr

4、inciples6uphold high standards of scientific excellence7 be made available for uses that accord with these principlesprimary purpose and usenature and uniquenessscalenature of Googles involvementProduct&use case reviews Identifying,assessing,and mitigating potential impacts before they are generally

5、 available to customers.Education,research&best practicesEquipping our customers with recommendations,thought leadership,and transparency as they navigate Responsible AI.RAI Tooling,Enablement,and Support Enabling our customers to identify,assess,and mitigate potential impacts within their use case and application(s).Google Clouds approach to Responsible AIResponsible AI Tooling,Enablement,and SupportEnabling our customers to identify,assess,and mitigate potential impacts within their use case and application(s).Safety filters and moderation APICitations toolingGround