2017
Selekoa’s 83 records dated 2017, in date order, grouped by topic.
Selective: what the records hold, not a history of the year.
How this page is made
- Each record keeps the kind of date its source gives: an event, a publication, an announcement, an effective date, a revision, a signature, a snapshot, or the start or end of an interval. A snapshot is one dated report of a state.
- A record dated only to a month sits under that month, before the records dated to a day. A record dated only to the year sits under the year, before the months. No day or month is added.
- An announcement records what a party stated. It does not show that anything was done, built or paid for.
- Dates a party states for a year are listed apart, after the records: targets, deadlines and terms, and dates stated in passing, such as a reported start of operation. So is a record’s own date that falls after its sources were last checked, such as a law’s date of effect. A stated date given only in words, with no year, is not listed.
- Records with an unknown date are not listed. A record with an interval appears in the year it starts and in the year it ends, or once, with both dates, when both fall in one year.
- Each provision of a text is a record of its own. Provisions sharing a text and a date are shown here as one entry, with their number, and listed on the text’s page; a contested or unresolved one is listed here on its own. The counts on this page include each provision record.
- The page is generated from the records, with no selection or framing by an editor. Coverage is uneven across years and topics: a year or topic with few records may be one the records barely cover.
People 15 records
Dated to 2017 only
· interval ends · dated to the year
Andrej Karpathy’s career timeline places his OpenAI research-scientist and founding-member work in 2015–2017. Evidence ↗· interval begins · dated to the year
Andrej Karpathy’s career timeline dates his Tesla Director of AI role to 2017–2022. Evidence ↗· snapshot · dated to the year
Daniel Levy’s biography places a Google Brain internship in 2017. Evidence ↗
June 2017
· snapshot · dated to the day
Attention Is All You Need v1 lists Ashish Vaswani with a Google Brain affiliation. Evidence ↗· snapshot · dated to the day
The first version of Deep reinforcement learning from human preferences lists Paul F Christiano with an OpenAI affiliation. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Łukasz Kaiser with a Google Brain affiliation. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Niki Parmar with a Google Research affiliation. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Noam Shazeer with a Google Brain affiliation. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Jakob Uszkoreit with a Google Research affiliation. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Aidan N. Gomez with a University of Toronto affiliation and a footnote that his work was performed while at Google Brain. Evidence ↗· snapshot · dated to the day
Attention Is All You Need (arXiv v1, 12 June 2017) lists Llion Jones under the Google Research affiliation. Evidence ↗
September 2017
· interval begins · dated to the month
MiniMax’s Hong Kong prospectus of 31 December 2025 states that Yun Yeyi held positions at SenseTime from September 2017 to January 2022, latterly as director of innovative business. Evidence ↗· snapshot · dated to the day
Google Brain’s September 2017 research account identifies Chris Olah as a Brain team member. Evidence ↗
October 2017
December 2017
Institutions 2 records
Dated to 2017 only
September 2017
· snapshot · dated to the day
Google Brain’s 2017 research account describes the team as part of Google Research. Evidence ↗
Ideas & lineages 64 records
Dated to 2017 only
· publication · dated to the year
“Ovatko hyvät johtamismenetelmät talouskasvun lähde? (pääkirjoitus)” (2017) credits Ari Hyytinen as its author. Evidence ↗· publication · dated to the year
Hyytinen, “Ovatko hyvät johtamismenetelmät talouskasvun lähde? (pääkirjoitus)” (2017) bears on E01. Hyytinen defines the "economic-growth singularity". Evidence ↗Hyytinen, Ovatko hyvät johtamismenetelmät talouskasvun lähde (2017) →Claim page →
· publication · dated to the year
Hyytinen, “Ovatko hyvät johtamismenetelmät talouskasvun lähde? (pääkirjoitus)” (2017) bears on E05. Editorial in the Finnish economics association's journal; it calls the idea "apparently improbable but interesting" before turning to management practices. Evidence ↗Hyytinen, Ovatko hyvät johtamismenetelmät talouskasvun lähde (2017) →Claim page →
· publication · dated to the year
Mullainathan and Spiess, “Machine Learning: An Applied Econometric Approach” (2017) bears on M04. Mullainathan and Spiess characterise machine learning as a prediction tool distinct from econometric parameter estimation, for use in empirical economics. Evidence ↗Mullainathan and Spiess, Machine Learning (2017) →Claim page →
June 2017
· event · dated to the day
DeepMind published its account of learning reward predictors from human feedback. Evidence ↗· event · dated to the day
DeepMind’s human-feedback account links Deep reinforcement learning from human preferences. Evidence ↗Learning through human feedback — DeepMind account (2017) →Claim page →
· event · dated to the day
DeepMind’s human-feedback account credits its research collaboration with OpenAI. Evidence ↗Learning through human feedback — DeepMind account (2017) →Claim page →
· event · dated to the day
Ashish Vaswani is a credited co-author of Attention Is All You Need v1. Evidence ↗· publication · dated to the day
Attention Is All You Need cites the Adam paper. Evidence ↗· publication · dated to the day
Attention Is All You Need reports using Adam to train its models. Evidence ↗· publication · dated to the day
Deep reinforcement learning from human preferences names Dario Amodei as a co-author. Evidence ↗· publication · dated to the day
Deep reinforcement learning from human preferences names Jan Leike as a co-author. Evidence ↗· publication · dated to the day
Human Preferences cites TRPO in its policy-optimisation procedure. Evidence ↗Deep reinforcement learning from human preferences →Claim page →
· publication · dated to the day
Human Preferences reports using TRPO for simulated robotics tasks. Evidence ↗Deep reinforcement learning from human preferences →Claim page →
· publication · dated to the day
Deep reinforcement learning from human preferences names Shane Legg as a co-author. Evidence ↗· event · dated to the day
OpenAI published its account of learning reward functions from human comparisons. Evidence ↗· event · dated to the day
OpenAI’s human-preferences account credits collaboration with DeepMind’s safety team. Evidence ↗Learning from human preferences — OpenAI account (2017) →Claim page →
July 2017
· publication · dated to the day
PPO includes an explicit reference to the Adam paper. Evidence ↗· publication · dated to the day
PPO cites Trust Region Policy Optimization in its discussion of trust-region methods. Evidence ↗· publication · dated to the day
PPO reports Adam step sizes for its MuJoCo, Roboschool and Atari experiments. Evidence ↗· publication · dated to the day
Proximal Policy Optimization Algorithms names John Schulman as a co-author. Evidence ↗
October 2017
· publication · dated to the month
“Artificial Intelligence and Economic Growth” (2017-10) credits Benjamin F. Jones as one of its 3 authors. Evidence ↗· publication · dated to the month
“Artificial Intelligence and Economic Growth” (2017-10) credits Charles I. Jones as one of its 3 authors. Evidence ↗· publication · dated to the month
“Artificial Intelligence and Economic Growth” (2017-10) credits Philippe Aghion as one of its 3 authors. Evidence ↗· publication · dated to the month
Aghion, Jones and Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates E01. The authors model "a unit of A.I. working to improve itself" by a differential equation for cognitive productivity. The authors cite Good (1965), Vinge (1993) and Kurzweil (2005) and point to Nordhaus (2015) for an economic discussion. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones and Jones, “Artificial Intelligence and Economic Growth” (2017-10) contests E01. The authors list bottlenecks that could prevent a singularity even with complete automation. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones and Jones, “Artificial Intelligence and Economic Growth” (2017-10) bears on E02. The authors note that their "labor free" singularities need complete automation of cognitive tasks. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones and Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates E05. The abstract lists singularities and superintelligence among the questions for an economics research agenda. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates G01. With every task in goods production automated, the model becomes an AK model and ongoing productivity growth makes growth rates themselves rise over time ("Type I": growth rates rising without bound). Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates G02. If an AI replaces all people in producing ideas and ideas build on earlier ideas, the model reaches infinite income in finite time. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) contests G02. If ideas become harder to find fast enough, even fully automated research does not produce a singularity. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates G04. The paper's organising theme, developed in its task-based models of automation. States the condition directly. The same example then shows that with Cobb-Douglas production a singularity can occur without full automation. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates G05. Gives physical limits, such as Carnot's theorem for energy conversion, as examples of essential inputs whose productivity cannot keep rising. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) bears on M01. Places AI in a continuous history of automation from the steam engine to semiconductors, while noting AI may now reach non-routine cognitive tasks. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates M02. Frames AI within a task-based automation model (building on Zeira 1998 and Acemoglu and Restrepo) in which tasks move from labour to capital. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) articulates M05. Models the case in which A.I. replaces people in all tasks to show how growth could explode; the authors then list objections (automation limits, search limits, Baumol bottlenecks). Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) bears on M05. Treats complete automation as an open, debated question and names incomplete automation as a bottleneck to explosive growth. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) bears on M06a. Argues that Baumol's cost disease can keep the labour share substantial even as automation advances. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the month
Aghion, Jones, Jones, “Artificial Intelligence and Economic Growth” (2017-10) bears on M09. Presents the paper as agenda-setting for research on AI and growth. Evidence ↗Aghion, Jones, Jones, Artificial Intelligence and Economic Growth (2017) →Claim page →
· publication · dated to the day
“There's No Fire Alarm for Artificial General Intelligence” (2017-10-13) credits Eliezer Yudkowsky as its author. Evidence ↗· publication · dated to the day
The post “There's No Fire Alarm for Artificial General Intelligence” (2017-10-13) contests Y08. Argues that, as with flight and nuclear fission, no event before AGI itself will create common knowledge that it is time to act, so waiting for such a signal means not acting. Evidence ↗Yudkowsky, There's No Fire Alarm for Artificial General Intelligence (2017) →Claim page →
· publication · dated to the day
The post “There's No Fire Alarm for Artificial General Intelligence” (2017-10-13) contests Y13. His second reason why apparent distance does not license inaction; he then gives the Wright brothers' and Fermi's own pessimistic forecasts as examples. Argues that people who defer alignment work until an unspecified future signal are unlikely to respond competently when a signal comes. Evidence ↗Yudkowsky, There's No Fire Alarm for Artificial General Intelligence (2017) →Claim page →
· event · dated to the day
The AlphaGo Zero account describes reinforcement learning through repeated self-play and search. Evidence ↗· event · dated to the day
DeepMind published its AlphaGo Zero account of learning Go through self-play. Evidence ↗· event · dated to the day
David Silver is credited as a co-author of the AlphaGo Zero research account. Evidence ↗
November 2017
· publication · dated to the month
“Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) credits Chad Syverson as one of its 3 authors. Evidence ↗· publication · dated to the month
“Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) credits Daniel Rock as one of its 3 authors. Evidence ↗· publication · dated to the month
“Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) credits Erik Brynjolfsson as one of its 3 authors. Evidence ↗· publication · dated to the month
Brynjolfsson, Rock, Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) articulates M03. Argues that implementation lags, rather than false hopes, mismeasurement or redistribution, are the likeliest main explanation of the gap between AI expectations and slowing measured productivity. Concludes that AI breakthroughs have not yet diffused and their complementary changes depress measured productivity at first; the same conclusion calls AI the core of "a new, economically important potential GPT". Evidence ↗Brynjolfsson, Rock, Syverson, Artificial Intelligence and the Modern Productivity Paradox (2017) →Claim page →
· publication · dated to the month
Brynjolfsson, Rock, Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) bears on M03. Formalises how unmeasured intangible AI-complementary investment first depresses and later inflates measured productivity growth. Evidence ↗Brynjolfsson, Rock, Syverson, Artificial Intelligence and the Modern Productivity Paradox (2017) →Claim page →
· publication · dated to the month
Brynjolfsson, Rock, Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) bears on M08. Illustrates that a single AI application could add about 0.17 percentage points a year to US labour productivity growth over a decade. Second worked example showing that individual AI applications could add measurably to aggregate labour productivity over about a decade. Evidence ↗Brynjolfsson, Rock, Syverson, Artificial Intelligence and the Modern Productivity Paradox (2017) →Claim page →
· publication · dated to the month
Brynjolfsson, Rock, Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics” (2017-11) bears on M09. Contrasts technologist optimism with the productivity pessimism the authors attribute to economists and statisticians, and argues both can be right. Evidence ↗Brynjolfsson, Rock, Syverson, Artificial Intelligence and the Modern Productivity Paradox (2017) →Claim page →
December 2017
· publication · dated to the month
“Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) credits Anton Korinek as one of its 2 authors. Evidence ↗· publication · dated to the month
“Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) credits Joseph E. Stiglitz as one of its 2 authors. Evidence ↗· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates E02. Korinek and Stiglitz open their speculative long-run section with expert forecasts (citing Bostrom 2014). Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates E03. The authors discuss whether humans can keep property rights over AI entities. Drawing on Korinek (2017), the authors describe a Malthusian race between humans and autonomous AI entities. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) bears on E03. The footnote qualifies the authors' assumption that "our social and economic system will be maintained" after AGI. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) bears on E04. The authors consider halting AI progress before turning to a human "reservation" policy. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates E05. The authors justify taking up AGI in an economics paper. The authors flag the speculative nature of their AI-entity scenario. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) bears on M03. One of three possibilities raised against the slow recent productivity record, alongside mismeasurement and a singularity-type discontinuity. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates M05. Describes a transition in which falling machine costs make machines a cheaper substitute for all human labour; with frictionless capital adjustment the human wage may be unchanged while the labour share goes to zero. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates M06a. Concludes that without redistribution, AI and other worker-replacing change can lower both labour's share and workers' absolute welfare. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) articulates M06b. Ties the case for redistribution to the distributional losses of worker-replacing technological change. Argues that policy, including non-distortionary taxation, can compensate those who lose from AI, even in a singularity case. Names patent length and capital taxation as second-best redistribution instruments for losses from worker-replacing machines. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
· publication · dated to the month
Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) bears on M09. Argues economists must say how gains from AI can be shared, not merely that they could be. Evidence ↗Korinek and Stiglitz, Artificial Intelligence and Its Implications for Income Distribution and Unemployment (2017) →Claim page →
Governance 2 records
October 2017
· event · dated to the day
DeepMind published the launch rationale for its Ethics & Society research unit. Evidence ↗· snapshot · dated to the day
DeepMind described Ethics & Society as a new internal research unit. Evidence ↗