Difference between revisions of "Timeline of AI timelines"

From Timelines
Jump to: navigation, search
(→‎What the timeline is still missing)
(→‎What the timeline is still missing)
 
(40 intermediate revisions by 2 users not shown)
Line 11: Line 11:
  
 
{| class="sortable wikitable"
 
{| class="sortable wikitable"
−
! Year !! Month and date !! Event type !! Details
+
! Year of prediction!! Predicted year !! Predictor !! Details
 +
|-
 +
| 1950 || 2000 || || In his paper, often regarded as an AI manifesto, Alan Turing foresees a future where computers, by the year 2000, would possess the capability to respond to questions posed by humans in a manner indistinguishable from human responses.<ref>{{cite web |title=3 Things Alan Turing Never Imagined |url=https://www.cmswire.com/digital-experience/3-things-alan-turing-never-imagined/ |website=CMSWire.com |access-date=26 October 2023 |language=en}}</ref>
 +
|-
 +
| 1965 || 1985 || {{w|Herbert A. Simon}} || AI pioneer {{w|Herbert A. Simon}} writes: "machines will be capable, within twenty years, of doing any work a man can do."<ref>{{Harvnb|Simon|1965|p=96}} quoted in {{Harvnb|Crevier|1993|p=109}}</ref>
 +
|-
 +
| 1965 || 2000 || {{w|Irving John Good}} || {{w|Irving John Good}} predicts an ultraintelligent machine by 2000.
 +
|-
 +
| 1988 || 2010 || {{w|Hans Moravec}} || {{w|Hans Moravec}} writes:
 +
{{quote|Behold my book Mind Children. Within, I project that, in 2010 or thereabouts, we shall achieve strong AI. I am not calling it “Artificial General Intelligence” because this term will not be coined for another 15 years or so.<ref name="Biology"/>}}
 +
|-
 +
| 1997 || || || The term "artificial general intelligence" is probably first used by Mark Gubrud in a discussion of the implications of fully automated military production and operations.<ref>{{Harvnb|Gubrud|1997}}</ref>
 +
|-
 +
| 2001 || 2023, 2049, 2059 || {{w|Ray Kurzweil}} || {{w|Ray Kurzweil}} writes:
 +
{{quote|I have calculated that matching the intelligence of a human brain requires 2 * 10^16 ops/sec* and this will become available in a $1000 computer in 2023. 26 years after that, in 2049, a $1000 computer will have ten billion times more computing power than a human brain; and in 2059, that computer will cost one cent.<ref name="Biology">{{cite web |title=Biology-Inspired AGI Timelines: The Trick That Never Works |url=https://www.greaterwrong.com/posts/ax695frGJEzGxFBK4/biology-inspired-agi-timelines-the-trick-that-never-works#__2004_or_thereabouts__ |website=www.greaterwrong.com |access-date=16 July 2022}}</ref>}}
 +
|-
 +
| 2011 (January) || 2050 || || The Future of Humanity Institute conducts the Winter Intelligence Survey during their AGI impacts conference. Participants estimate a median likelihood of 50% for human-level AI development by the year 2050. The survey includes questions about the probabilities of achieving human-level machine intelligence by specific years. Responses variy, with a 10% chance estimated around 2015-2030, a 50% chance around 2040-2080, and a 90% chance around 2100-2250. The participants, primarily experts in AI and related fields, are highly interested in AI's impact, and their responses indicate a range of expectations regarding AI development timelines.<ref>{{cite web |title=FHI Winter Intelligence Survey |url=https://aiimpacts.org/fhi-ai-timelines-survey/ |website=AI Impacts |access-date=26 October 2023 |date=29 December 2014}}</ref><ref>{{cite web |title=Machine Intelligence Survey |url=https://www.fhi.ox.ac.uk/wp-content/uploads/2011-1.pdf |website=fhi |access-date=12 August 2022}}</ref>
 +
|-
 +
| 2011 (August) || <2030 || || The AGI-11 survey is conducted at the AGI-11 conference with 60 participants, revealing that nearly half of the respondents believe Artificial General Intelligence (AGI) would be achieved before 2030. Additionally, almost 90% anticipate AGI appearing before 2100, and around 85% believe it would benefit humanity. The survey, consisting of two questions, is conducted by James Barrat and Ben Goertzel. Despite some concerns about the survey structure expressed in comments, the results are consistent with a majority expecting AGI to be implemented in the near future and to have positive implications for humankind.<ref>{{cite web |title=AGI-11 survey |url=https://aiimpacts.org/agi-11-survey/ |website=AI Impacts |access-date=26 October 2023 |date=10 November 2018}}</ref><ref>{{cite web |title=The Fourth Conference on Artificial General Intelligence |url=https://agi-conference.org/2011/ |website=agi-conference.org |access-date=26 October 2023}}</ref>
 +
|-
 +
| 2012–2013 || 2040 || || Vincent Müller and Nick Bostrom from FHI conduct a survey involving four different sets of AI experts between 2012 and 2013. Collectively, the experts estimate a 10% probability of achieving human-level AI by 2022 and a 50% probability by 2040.<ref>{{cite web |title=Müller and Bostrom AI Progress Poll |url=https://aiimpacts.org/muller-and-bostrom-ai-progress-poll/ |website=AI Impacts |access-date=12 August 2022 |date=29 December 2014}}</ref>
 +
|-
 +
| 2016 (March) || >2041 || {{w|Association for the Advancement of Artificial Intelligence}} fellows || Oren Etzioni conducts a survey among 193 AAAI fellows, prominent figures in the field of artificial intelligence. The survey asks about the achievement of 'Superintelligence,' defined as intellect surpassing human capabilities in various fields. Out of the 80 responses (a 41% response rate), 67.5% believe that Superintelligence would be achieved, but in more than 25 years. None expects it within the next 10 years, 7.5% think it might happen in the next 10-25 years, and 25% believes it would never occur.<ref>{{cite web |title=The AAAI Fellows Program |url=https://aaai.org/about-aaai/aaai-awards/the-aaai-fellows-program/ |website=AAAI |access-date=26 October 2023}}</ref>
 +
|-
 +
| 2016 (June 9) || || || {{w|Bill Gates}} says that achieving “human-level” AI will take “at least 5 times as long as what Ray Kurzweil says.<ref>{{cite web |title=Bill Gates on AI timelines |url=https://lukemuehlhauser.com/bill-gates-on-ai-timelines/ |website=lukemuehlhauser.com |access-date=12 August 2022}}</ref> 
 +
|-
 +
| 2017 (January–February) || 2026 || || Professor Toby Walsh conducts a survey involving 849 participants, including AI experts, robotics experts, and non-experts. The respondents classify 70 occupations as either at risk or not of automation within the next two decades. The survey uses machine learning rankings for these occupations. Additionally, participants predict the arrival of high-level machine intelligence (HLMI) - defined as computers performing human professions as well as humans - by selecting specific years with different probabilities. Median responses indicate varying estimates among the groups for the potential automation of occupations and the timeline for HLMI arrival. Non-Experts estimate a 10% probability of HLMI by 2026.<ref name="Walsh 2017 survey">{{cite web |title=Walsh 2017 survey |url=https://aiimpacts.org/walsh-2017-survey/ |website=AI Impacts |access-date=26 October 2023 |date=24 December 2019}}</ref> 
 +
|-
 +
| 2017 (January–February) || 2060 || || Non-Experts estimate a 90% probability of HLMI by 2060 in the Walsh 2017 survey. The group of non-experts includes 548 individuals who read an article about AI on The Conversation website. Although specific data about their expertise in AI or robotics is not gathered, Toby Walsh indicates that it is safe to assume that the majority are not experts in the AI or robotics fields and are unlikely to publish in prominent conferences like IJCAI, AAAI, or ICRA.<ref name="Walsh 2017 survey"/>
 +
|-
 +
| 2017 (January–February) || 2109 || || AI Experts estimate a 90% probability of HLMI by 2109.<ref name="Walsh 2017 survey"/>
 +
|-
 +
| 2017 (January–February) || 2118 || Electrical and Electronics Engineers (IEEE) Robotics & Automation Society fellows || Robotics experts estimate a 90% probability of HLMI by 2118. The group of robotics experts comprises 101 members, all of whom are either distinguished Fellows of the Institute for Electrical and Electronics Engineers (IEEE) Robotics & Automation Society or contributors to the 2016 IEEE Conference on Robotics & Automation (ICRA).<ref name="Walsh 2017 survey"/>
 +
|-
 +
| 2017 || || || A study finds 50% of AI researchers accord a 50% probability to "High level machine intelligence" (HLMI) by 2040, while 20% say that 50% probability will not be reached until 2100 or later. The survey finds an aggregated probability distribution with a 25%-75% confidence interval ranging from 2040 to well past 2100.<ref name="metaculus">{{cite web |title=When will the first weakly general AI system be devised, tested, and publicly announced? |url=https://www.metaculus.com/questions/3479/date-weakly-general-ai-is-publicly-known/ |website=www.metaculus.com |access-date=12 August 2022 |date=18 January 2020}}</ref><ref>{{cite web |title=Expert and Non-Expert Opinion about Technological Unemployment |url=https://arxiv.org/ftp/arxiv/papers/1706/1706.06906.pdf |website=arxiv.org |access-date=12 August 2022}}</ref>
 +
|-
 +
| 2021 (August 17) || 2036, 2060, 2100 || {{w|Holden Karnofsky}} || {{w|Holden Karnofsky}} predicts: "there's more than a 10% chance of ... PASTA-like "transformative AI" within 15 years (by 2036); a ~50% chance we'll see it within 40 years (by 2060); and a ~2/3 chance we'll see it this century (by 2100)."<ref>{{cite web |title=A public prediction by Holden Karnofsky |url=https://www.metaculus.com/prediction/8978/a-public-prediction-by-holden-karnofsky/ |website=www.metaculus.com |access-date=12 August 2022 |date=2 January 2022}}</ref>
 +
|-
 +
| 2022 (May 30) || 2029 || {{w|Elon Musk}} || {{w|Elon Musk}} tells {{w|Jack Dorsey}} via tweet that AGI would be achieved by 2029.<ref>{{cite web |title=https://twitter.com/elonmusk/status/1531328534169493506 |url=https://twitter.com/elonmusk/status/1531328534169493506 |website=Twitter |access-date=12 August 2022 |language=en}}</ref> {{w|Gary Marcus}} would criticize this, challenging Musk to a $100,000 bet.<ref>{{cite web |last1=Daws |first1=Ryan |title=Gary Marcus criticises Elon Musk’s AGI prediction |url=https://www.artificialintelligence-news.com/2022/06/01/gary-marcus-criticises-elon-musk-agi-prediction/#:~:text=Gary%20Marcus%20has%20criticised%20a,him%20to%20a%20%24100%2C000%20bet. |website=AI News |access-date=12 August 2022 |date=1 June 2022}}</ref>
 +
|-
 +
| 2022 (June–August) || 2059 || || The 2022 Expert Survey on Progress in AI (ESPAI) conducted by AI Impacts analyzes responses from 738 machine learning researchers. The survey aims to predict advancements in high-level machine intelligence (HLMI). Based on the data, the median prediction for a 50% chance of HLMI is 37 years, approximately in 2059. This represents a decrease of about eight years from 2016's prediction, with the caveat that human scientific activity must continue without major disruption. Concerning the impact of advanced AI on humanity, the median respondent believes there is a 5% chance of an extremely negative outcome, such as human extinction. Additionally, 69% of participants indicate that society should prioritize AI safety research more than current efforts. Most respondents (54%) think there is an "about even chance" that arguments for an intelligence explosion are correct.<ref>{{cite web |title=2022 Expert Survey on Progress in AI |url=https://aiimpacts.org/2022-expert-survey-on-progress-in-ai/#Results |website=AI Impacts |access-date=12 August 2022 |date=4 August 2022}}</ref>
 +
|-
 +
| 2023 (November 29) || 2026 || {{w|Elon Musk}} || In an interview, {{w|Elon Musk}} provides a three-year timeline for when artificial intelligence (AI) would surpass the intelligence of the smartest human.<ref>{{cite web |last1=Kastrenakes |first1=Jacob |title=Musk thinks we’re three years from super intelligent AI. |url=https://www.theverge.com/2023/11/29/23981844/musk-thinks-were-three-years-from-super-intelligent-ai |website=The Verge |access-date=20 April 2024 |language=en |date=29 November 2023}}</ref>
 
|-
 
|-
 
|}
 
|}
Line 19: Line 61:
 
===How the timeline was built===
 
===How the timeline was built===
  
−
The initial version of the timeline was written by [[User:FIXME|FIXME]].
+
The initial version of the timeline was written by [[User:Sebastian|Sebastian]].
  
 
{{funding info}} is available.
 
{{funding info}} is available.
Line 31: Line 73:
 
===What the timeline is still missing===
 
===What the timeline is still missing===
  
−
* probably go back to people like Turing, IJ Good, or whoever who first came up with ideas around AGI; what did they say about when it would be created?
+
* {{w|Progress in artificial intelligence}}
 +
* Sebastian: https://time.com/6556168/when-ai-outsmart-humans/
 +
* Sebastian: https://medium.com/accelerated-intelligence/top-ai-experts-predict-artificial-superintelligence-in-3-5-years-now-what-fbfe00204c7f
 +
* Sebastian: https://research.aimultiple.com/artificial-general-intelligence-singularity-timing/
 +
* Sebastian: https://www.theguardian.com/technology/2024/feb/17/humanitys-remaining-timeline-it-looks-more-like-five-years-than-50-meet-the-neo-luddites-warning-of-an-ai-apocalypse
 +
* Sebastian: https://techcrunch.com/2024/03/19/agi-and-hallucinations/?fbclid=IwAR2_TayMg3U_HrIk21tXokSeTiOAkvnF5pU8KjkS9b_AHoVcQVOpxTnCH6g
 +
* probably go back to people like Turing✔, IJ Good, or whoever who first came up with ideas around AGI; what did they say about when it would be created?
 
* dartmouth conference and initial optimistic ideas on when AGI would be created
 
* dartmouth conference and initial optimistic ideas on when AGI would be created
 
* would be cool to include the various "standalone"/"self-contained" AI timeline methods and when they were first used https://wiki.issarice.com/wiki/Category:AI_timelines_arguments
 
* would be cool to include the various "standalone"/"self-contained" AI timeline methods and when they were first used https://wiki.issarice.com/wiki/Category:AI_timelines_arguments
Line 52: Line 100:
 
* the AI Impacts experts survey paper from 2016 https://arxiv.org/pdf/1705.08807.pdf
 
* the AI Impacts experts survey paper from 2016 https://arxiv.org/pdf/1705.08807.pdf
 
* look through posts at https://aiimpacts.org/category/ai-timelines/
 
* look through posts at https://aiimpacts.org/category/ai-timelines/
 +
 +
 +
 +
* https://thenextweb.com/news/what-is-the-technological-singularity#.tnw_4ol0E0Vr
 +
* https://voicebot.ai/2023/08/16/generative-ai-revenue-will-reach-1-3t-in-2032-chart/
 +
* https://www.knowyourmobile.com/ai-intelligence/ai-2033-a-glimpse-into-the-future-of-artificial-intelligence/ (2033)
  
 
===Timeline update strategy===
 
===Timeline update strategy===

Latest revision as of 12:19, 28 August 2024

This is a timeline of AI timelines, the study of advances in artificial intelligence, in particular when artificial general intelligence will be created.

Big picture

Time period Development summary More details

Full timeline

Year of prediction Predicted year Predictor Details
1950 2000 In his paper, often regarded as an AI manifesto, Alan Turing foresees a future where computers, by the year 2000, would possess the capability to respond to questions posed by humans in a manner indistinguishable from human responses.[1]
1965 1985 Herbert A. Simon AI pioneer Herbert A. Simon writes: "machines will be capable, within twenty years, of doing any work a man can do."[2]
1965 2000 Irving John Good Irving John Good predicts an ultraintelligent machine by 2000.
1988 2010 Hans Moravec Hans Moravec writes:
Behold my book Mind Children. Within, I project that, in 2010 or thereabouts, we shall achieve strong AI. I am not calling it “Artificial General Intelligence” because this term will not be coined for another 15 years or so.[3]
1997 The term "artificial general intelligence" is probably first used by Mark Gubrud in a discussion of the implications of fully automated military production and operations.[4]
2001 2023, 2049, 2059 Ray Kurzweil Ray Kurzweil writes:
I have calculated that matching the intelligence of a human brain requires 2 * 10^16 ops/sec* and this will become available in a $1000 computer in 2023. 26 years after that, in 2049, a $1000 computer will have ten billion times more computing power than a human brain; and in 2059, that computer will cost one cent.[3]
2011 (January) 2050 The Future of Humanity Institute conducts the Winter Intelligence Survey during their AGI impacts conference. Participants estimate a median likelihood of 50% for human-level AI development by the year 2050. The survey includes questions about the probabilities of achieving human-level machine intelligence by specific years. Responses variy, with a 10% chance estimated around 2015-2030, a 50% chance around 2040-2080, and a 90% chance around 2100-2250. The participants, primarily experts in AI and related fields, are highly interested in AI's impact, and their responses indicate a range of expectations regarding AI development timelines.[5][6]
2011 (August) <2030 The AGI-11 survey is conducted at the AGI-11 conference with 60 participants, revealing that nearly half of the respondents believe Artificial General Intelligence (AGI) would be achieved before 2030. Additionally, almost 90% anticipate AGI appearing before 2100, and around 85% believe it would benefit humanity. The survey, consisting of two questions, is conducted by James Barrat and Ben Goertzel. Despite some concerns about the survey structure expressed in comments, the results are consistent with a majority expecting AGI to be implemented in the near future and to have positive implications for humankind.[7][8]
2012–2013 2040 Vincent Müller and Nick Bostrom from FHI conduct a survey involving four different sets of AI experts between 2012 and 2013. Collectively, the experts estimate a 10% probability of achieving human-level AI by 2022 and a 50% probability by 2040.[9]
2016 (March) >2041 Association for the Advancement of Artificial Intelligence fellows Oren Etzioni conducts a survey among 193 AAAI fellows, prominent figures in the field of artificial intelligence. The survey asks about the achievement of 'Superintelligence,' defined as intellect surpassing human capabilities in various fields. Out of the 80 responses (a 41% response rate), 67.5% believe that Superintelligence would be achieved, but in more than 25 years. None expects it within the next 10 years, 7.5% think it might happen in the next 10-25 years, and 25% believes it would never occur.[10]
2016 (June 9) Bill Gates says that achieving “human-level” AI will take “at least 5 times as long as what Ray Kurzweil says.[11]
2017 (January–February) 2026 Professor Toby Walsh conducts a survey involving 849 participants, including AI experts, robotics experts, and non-experts. The respondents classify 70 occupations as either at risk or not of automation within the next two decades. The survey uses machine learning rankings for these occupations. Additionally, participants predict the arrival of high-level machine intelligence (HLMI) - defined as computers performing human professions as well as humans - by selecting specific years with different probabilities. Median responses indicate varying estimates among the groups for the potential automation of occupations and the timeline for HLMI arrival. Non-Experts estimate a 10% probability of HLMI by 2026.[12]
2017 (January–February) 2060 Non-Experts estimate a 90% probability of HLMI by 2060 in the Walsh 2017 survey. The group of non-experts includes 548 individuals who read an article about AI on The Conversation website. Although specific data about their expertise in AI or robotics is not gathered, Toby Walsh indicates that it is safe to assume that the majority are not experts in the AI or robotics fields and are unlikely to publish in prominent conferences like IJCAI, AAAI, or ICRA.[12]
2017 (January–February) 2109 AI Experts estimate a 90% probability of HLMI by 2109.[12]
2017 (January–February) 2118 Electrical and Electronics Engineers (IEEE) Robotics & Automation Society fellows Robotics experts estimate a 90% probability of HLMI by 2118. The group of robotics experts comprises 101 members, all of whom are either distinguished Fellows of the Institute for Electrical and Electronics Engineers (IEEE) Robotics & Automation Society or contributors to the 2016 IEEE Conference on Robotics & Automation (ICRA).[12]
2017 A study finds 50% of AI researchers accord a 50% probability to "High level machine intelligence" (HLMI) by 2040, while 20% say that 50% probability will not be reached until 2100 or later. The survey finds an aggregated probability distribution with a 25%-75% confidence interval ranging from 2040 to well past 2100.[13][14]
2021 (August 17) 2036, 2060, 2100 Holden Karnofsky Holden Karnofsky predicts: "there's more than a 10% chance of ... PASTA-like "transformative AI" within 15 years (by 2036); a ~50% chance we'll see it within 40 years (by 2060); and a ~2/3 chance we'll see it this century (by 2100)."[15]
2022 (May 30) 2029 Elon Musk Elon Musk tells Jack Dorsey via tweet that AGI would be achieved by 2029.[16] Gary Marcus would criticize this, challenging Musk to a $100,000 bet.[17]
2022 (June–August) 2059 The 2022 Expert Survey on Progress in AI (ESPAI) conducted by AI Impacts analyzes responses from 738 machine learning researchers. The survey aims to predict advancements in high-level machine intelligence (HLMI). Based on the data, the median prediction for a 50% chance of HLMI is 37 years, approximately in 2059. This represents a decrease of about eight years from 2016's prediction, with the caveat that human scientific activity must continue without major disruption. Concerning the impact of advanced AI on humanity, the median respondent believes there is a 5% chance of an extremely negative outcome, such as human extinction. Additionally, 69% of participants indicate that society should prioritize AI safety research more than current efforts. Most respondents (54%) think there is an "about even chance" that arguments for an intelligence explosion are correct.[18]
2023 (November 29) 2026 Elon Musk In an interview, Elon Musk provides a three-year timeline for when artificial intelligence (AI) would surpass the intelligence of the smartest human.[19]

Meta information on the timeline

How the timeline was built

The initial version of the timeline was written by Sebastian.

Funding information for this timeline is available.

Feedback and comments

Feedback for the timeline can be provided at the following places:

  • FIXME

What the timeline is still missing


Timeline update strategy

See also

External links

References

  1. ↑ "3 Things Alan Turing Never Imagined". CMSWire.com. Retrieved 26 October 2023. 
  2. ↑ Simon 1965, p. 96 quoted in Crevier 1993, p. 109
  3. ↑ 3.0 3.1 "Biology-Inspired AGI Timelines: The Trick That Never Works". www.greaterwrong.com. Retrieved 16 July 2022. 
  4. ↑ Gubrud 1997
  5. ↑ "FHI Winter Intelligence Survey". AI Impacts. 29 December 2014. Retrieved 26 October 2023. 
  6. ↑ "Machine Intelligence Survey" (PDF). fhi. Retrieved 12 August 2022. 
  7. ↑ "AGI-11 survey". AI Impacts. 10 November 2018. Retrieved 26 October 2023. 
  8. ↑ "The Fourth Conference on Artificial General Intelligence". agi-conference.org. Retrieved 26 October 2023. 
  9. ↑ "Müller and Bostrom AI Progress Poll". AI Impacts. 29 December 2014. Retrieved 12 August 2022. 
  10. ↑ "The AAAI Fellows Program". AAAI. Retrieved 26 October 2023. 
  11. ↑ "Bill Gates on AI timelines". lukemuehlhauser.com. Retrieved 12 August 2022. 
  12. ↑ 12.0 12.1 12.2 12.3 "Walsh 2017 survey". AI Impacts. 24 December 2019. Retrieved 26 October 2023. 
  13. ↑ "When will the first weakly general AI system be devised, tested, and publicly announced?". www.metaculus.com. 18 January 2020. Retrieved 12 August 2022. 
  14. ↑ "Expert and Non-Expert Opinion about Technological Unemployment" (PDF). arxiv.org. Retrieved 12 August 2022. 
  15. ↑ "A public prediction by Holden Karnofsky". www.metaculus.com. 2 January 2022. Retrieved 12 August 2022. 
  16. ↑ "https://twitter.com/elonmusk/status/1531328534169493506". Twitter. Retrieved 12 August 2022.  External link in |title= (help)
  17. ↑ Daws, Ryan (1 June 2022). "Gary Marcus criticises Elon Musk's AGI prediction". AI News. Retrieved 12 August 2022. 
  18. ↑ "2022 Expert Survey on Progress in AI". AI Impacts. 4 August 2022. Retrieved 12 August 2022. 
  19. ↑ Kastrenakes, Jacob (29 November 2023). "Musk thinks we're three years from super intelligent AI.". The Verge. Retrieved 20 April 2024. 
Retrieved from "http://timelines.issarice.com/index.php?title=Timeline_of_AI_timelines&oldid=81553"