牛津大学：AI+超越人类编年史.pptx

When Will AI Exceed Human Performance? Evidence from AI ExpertsKatja Grace1,2, John Salvatier2, Allan Dafoe1,3, Baobao Zhang3, and Owain Evans11Future of Humanity Institute, Oxford University2AI Impacts3Department of Political Science, Yale UniversityAbstractAdvances in artificial intelligence (AI) will transform modern life by reshaping transportation, health, science, finance, and the military [1, 2, 3]. To adapt public policy, we need to better anticipate these advances [4, 5]. Here we report the results from a large survey of machine learning researchers on their beliefs about progress in AI. Researchers predict AI will outper- form humans in many activities in the next ten years, such as translating languages (by 2024), writing high-school essays (by 2026), driving a truck (by 2027), working in retail (by 2031), writing a bestselling book (by 2049), and working as a surgeon (by 2053). Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 years, with Asian respondents expecting these dates much sooner than North Americans. These results will inform discussion amongst researchers and policymakers about anticipating and managing trends in AI.IntroductionAdvances in artificial intelligence (AI) will have massive social consequences. Self-driving tech- nology might replace millions of driving jobs over the coming decade. In addition to possible unemployment, the transition will bring new challenges, such as rebuilding infrastructure, pro- tecting vehicle cyber-security, and adapting laws and regulations [5]. New challenges, both for AI developers and policy-makers, will also arise from applications in law enforcement, military tech- nology, and marketing [6]. To prepare for these challenges, accurate forecasting of transformative AI would be invaluable. Several sources provide objective evidence about future AI advances: trends in computing hardware [7], task performance [8], and the automation of labor [9]. The predictions of AI experts provide crucial additional information. We survey a larger and more representative sample of AI experts than any study to date [10, 11]. Our questions cover the timing of AI advances (including both practical applications of AI and the automation of various human jobs), as well as the social and ethical impacts of AI.Survey MethodOur survey population was all researchers who published at the 2015 NIPS and ICML confer- ences (two of the premier venues for peer-reviewed research in machine learning). A total of 352 researchers responded to our survey invitation (21% of the 1634 authors we contacted). Our ques- tions concerned the timing of specific AI capabilities (e.g. folding laundry, language translation), superiority at specific occupations (e.g. truck driver, surgeon), superiority over humans at all tasks, and the social impacts of advanced AI. See Survey Content for details.Time Until Machines Outperform HumansAI would have profound social consequences if all tasks were more cost effectively accomplished by machines. Our survey used the following definition:“High-level machine intelligence” (HLMI) is achieved when unaided machines can ac- complish every task better and more cheaply than human workers.1Each individual respondent estimated the probability of HLMI arriving in future years. Taking the mean over each individual, the aggregate forecast gave a 50% chance of HLMI occurring within 45 years and a 10% chance of it occurring within 9 years. Figure 1 displays the probabilistic predictions for a random subset of individuals, as well as the mean predictions. There is large inter-subject variation: Figure 3 shows that Asian respondents expect HLMI in 30 years, whereas North Americans expect it in 74 years.0.000.250.500.751.0002550 Years from 201675100Probability of HLMIAggregate Forecast (with 95% Confidence Interval) Random Subset of Individual ForecastsLOESSFigure 1: Aggregate subjective probability of ‘high-level machine intelligence’ arrival by future years. Each respondent provided three data points for their forecast and these were fit to the Gamma CDF by least squares to produce the grey CDFs. The “Aggregate Forecast” is the mean distribution over all individual CDFs (also called the “mixture” distribution). The confidence interval was generated by bootstrapping (clustering on respondents) and plotting the 95% interval for estimated probabilities at each year. The LOESS curve is a non-parametric regression on all data points.While most participants were asked about HLMI, a subset were asked a logically similar question that emphasized consequences for employment. The question defined full automation of labor as:when all occupations are。