Progress in artificial intelligence
Artificial intelligence applications have been used in a wide range of fields including medical diagnosis, stock trading, robot control, law, scientific discovery and toys. However, many AI applications are not perceived as AI: "A lot of cutting edge AI has filtered into general applications, often without being called AI because once something becomes useful enough and common enough it's not labeled AI anymore." "Many thousands of AI applications are deeply embedded in the infrastructure of every industry." In the late 1990s and early 21st century, AI technology became widely used as elements of larger systems, but the field is rarely credited for these successes.
To allow comparison with human performance, artificial intelligence can be evaluated on constrained and well-defined problems. Such tests have been termed subject matter expert Turing tests. Also, smaller problems provide more achievable goals and there are an ever-increasing number of positive results.
|Game||Champion year||Legal states (log10)||Game tree complexity (log10)||Game of perfect information?||Ref|
|2p no-limit hold 'em||2017||Imperfect|
In his famous Turing test, Alan Turing picked language, the defining feature of human beings, for its basis. Yet, there are many other useful abilities that can be described as showing some form of intelligence. This gives better insight into the comparative success of artificial intelligence in different areas.
In what has been called the Feigenbaum test, the inventor of expert systems argued for subject specific expert tests. A paper by Jim Gray of Microsoft in 2003 suggested extending the Turing test to speech understanding, speaking and recognizing objects and behavior.
AI, like electricity or the steam engine, is a general purpose technology. There is no consensus on how to characterize which tasks AI tends to excel at. Some versions of Moravec's paradox observe that humans are more likely to outperform machines in areas such as physical dexterity that have been the direct target of natural selection. While projects such as AlphaZero have succeeded in generating their own knowledge from scratch, many other machine learning projects require large training datasets. Researcher Andrew Ng has suggested, as a "highly imperfect rule of thumb", that "almost anything a typical human can do with less than one second of mental thought, we can probably now or in the near future automate using AI."
Games provide a high-profile benchmark for assessing rates of progress; many games have a large professional player base and a well-established competitive rating system. AlphaGo brought the era of classical board-game benchmarks to an close. Games of imperfect knowledge provide new challenges to AI in the area of game theory; the most prominent milestone in this area was brought to a close by Libratus' poker victory in 2017. E-sports continue to provide additional benchmarks; Facebook AI, Deepmind, and others have engaged with the popular StarCraft franchise of videogames.
Broad classes of outcome for an AI test may be given as:
- optimal: it is not possible to perform better (note: some of these entries were solved by humans)
- super-human: performs better than all humans
- high-human: performs better than most humans
- par-human: performs similarly to most humans
- sub-human: performs worse than most humans
- Connect Four: 1988
- Checkers (aka 8x8 draughts): Weakly solved (2007)
- Rubik's Cube: Mostly solved (2010)
- Heads-up limit hold'em poker: Statistically optimal in the sense that "a human lifetime of play is not sufficient to establish with statistical significance that the strategy is not an exact solution" (2015)
- Othello (aka reversi): c. 1997
- Scrabble: 2006
- Backgammon: c. 1995-2002
- Chess: Supercomputer (c. 1997); Personal computer (c. 2006); Mobile phone (c. 2009); Computer defeats human + computer (c. 2017)
- Jeopardy: Question answering, although the machine did not use speech recognition (2011)
- Shogi: c. 2017
- Arimaa: 2015
- Go: 2017
- Heads-up no-limit hold'em poker: 2017
- Optical character recognition for printed text (nearing par-human for Latin-script typewritten text)
- Object recognition
- Facial recognition: Low to mid human accuracy (as of 2014)
- Visual question answering, such as the VQA 1.0
- Various robotics tasks that may require advances in robot hardware as well as AI, including:
- Stable bipedal locomotion: Bipedal robots can walk, but are less stable than human walkers (as of 2017)
- Humanoid soccer
- Speech recognition: "nearly equal to human performance" (2017)
- Explainability. Current medical systems can diagnose certain medical conditions well, but cannot explain to users why they made the diagnosis.
- Various tasks that are difficult to solve without contextual knowledge, including:
Past and current predictions
An expert poll around 2016, conducted by Katja Grace of the Future of Humanity Institute and associates, gave median estimates of 3 years for championship Angry Birds, 4 years for the World Series of Poker, and 6 years for StarCraft. On more subjective tasks, the poll gave 6 years for folding laundry as well as an average human worker, 7–10 years for expertly answering 'easily Googleable' questions, 8 years for average speech transcription, 9 years for average telephone banking, and 11 years for expert songwriting, but over 30 years for writing a New York Times bestseller or winning the Putnam math competition.
|Year prediction made||Predicted year||Number of Years||Predictor||Contemporaneous source|
|1957||1967 or sooner||10 or less||Herbert A. Simon, economist|
|1990||2000 or sooner||10 or less||Ray Kurzweil, futurist||Age of Intelligent Machines|
AlphaGo defeated a European Go champion in October 2015, and defeated one of the world's top players, Lee Sedol, in March 2016. According to Scientific American and other sources, most observers had expected superhuman Computer Go performance to be at least a decade away.
|Year prediction made||Predicted year||Number of years||Predictor||Affiliation||Contemporaneous source|
|1997||2100 or later||100 or more||Piet Hutt, physicist and Go fan||Institute for Advanced Study||New York Times|
|2007||2017 or sooner||10 or less||Feng-Hsiung Hsu, Deep Blue lead||Microsoft Research Asia||IEEE Spectrum|
|2014||2024||10||Rémi Coulom, Computer Go programmer||CrazyStone||Wired|
Human-level artificial general intelligence (AGI)
AI pioneer and economist Herbert A. Simon inaccurately predicted in 1965: "Machines will be capable, within twenty years, of doing any work a man can do". Similarly, in 1970 Marvin Minsky wrote that "Within a generation... the problem of creating artificial intelligence will substantially be solved."
The Grace poll around 2016 found results varied depending on how the question was framed. Respondents asked to estimate "when unaided machines can accomplish every task better and more cheaply than human workers" gave an aggregated median answer of 45 years and a 10% chance of it occurring within 9 years. Other respondents asked to estimate "when all occupations are fully automatable. That is, when for any occupation, machines could be built to carry out the task better and more cheaply than human workers" estimated a median of 122 years and a 10% probability of 20 years. The median response for when "AI researcher" could be fully automated was around 90 years. No link was found between seniority and optimism, but Asian researchers were much more optimistic than North American researchers on average; Asians predicted 30 years on average for "accomplish every task", compared with the 74 years predicted by North Americans.
|Year prediction made||Predicted year||Number of years||Predictor||Contemporaneous source|
|1965||1985 or sooner||20 or less||Herbert A. Simon||The shape of automation for men and management|
|1993||2023 or sooner||30 or less||Vernor Vinge, science fiction writer||"The Coming Technological Singularity"|
|1995||2040 or sooner||45 or less||Hans Moravec, robotics researcher||Wired|
|2008||Never||Gordon E. Moore, inventor of Moore's Law||IEEE Spectrum|
- AI set to exceed human brain power CNN.com (July 26, 2006)
- Kurtzweil 2005, p. 264
- National Research Council (1999), "Developments in Artificial Intelligence", Funding a Revolution: Government Support for Computing Research, National Academy Press, ISBN 0-309-06278-0, OCLC 246584055 under "Artificial Intelligence in the 90s"
- Approximate year AI started beating top human experts
- van den Herik, H.Jaap; Uiterwijk, Jos W.H.M.; van Rijswijck, Jack (January 2002). "Games solved: Now and in the future". Artificial Intelligence. 134 (1–2): 277–311. doi:10.1016/S0004-3702(01)00152-7.
- "www.othello-club.de". berg.earthlingz.de. Retrieved 2018-07-15.
- Madrigal, Alexis C. (2017). "How Checkers Was Solved". The Atlantic. Retrieved 6 May 2018.
- Webley, Kayla (15 February 2011). "Top 10 Man-vs.-Machine Moments". Time. Retrieved 28 December 2017.
- "Shogi prodigy breathes new life into the game | The Japan Times". The Japan Times. Retrieved 2018-07-15.
- Brown, Noam; Sandholm, Tuomas (2017). "Superhuman AI for heads-up no-limit poker: Libratus beats top professionals". Science. 359 (6374): 418. doi:10.1126/science.aao1733. PMID 29249696.
- "Facebook Quietly Enters StarCraft War for AI Bots, and Loses". WIRED. 2017. Retrieved 6 May 2018.
- Turing, Alan (October 1950), "Computing Machinery and Intelligence", Mind, LIX (236): 433–460, doi:10.1093/mind/LIX.236.433, ISSN 0026-4423
- Feigenbaum, Edward A. (2003). "Some challenges and grand challenges for computational intelligence". Journal of the ACM. 50 (1): 32–40. doi:10.1145/602382.602400.
- Gray, Jim (2003). "What Next? A Dozen Information-Technology Research Goals". Journal of the ACM. 50 (1): 41–57. arXiv:cs/9911005
[cs.GL]. Bibcode:1999cs.......11005G. doi:10.1145/602382.602401.
- Brynjolfsson, Erik; Mitchell, Tom (22 December 2017). "What can machine learning do? Workforce implications". Science. pp. 1530–1534. Bibcode:2017Sci...358.1530B. doi:10.1126/science.aap8062. Retrieved 7 May 2018.
- "IKEA furniture and the limits of AI". The Economist. 2018. Retrieved 24 April 2018.
- Sample, Ian (18 October 2017). "'It's able to create knowledge itself': Google unveils AI that learns on its own". the Guardian. Retrieved 7 May 2018.
- "The AI revolution in science". Science | AAAS. 5 July 2017. Retrieved 7 May 2018.
- "Will your job still exist in 10 years when the robots arrive?". South China Morning Post. 2017. Retrieved 7 May 2018.
- Borowiec, Tracey Lien, Steven (2016). "AlphaGo beats human Go champ in milestone for artificial intelligence". latimes.com. Retrieved 7 May 2018.
- Brown, Noam; Sandholm, Tuomas (26 January 2018). "Superhuman AI for heads-up no-limit poker: Libratus beats top professionals". Science. pp. 418–424. doi:10.1126/science.aao1733. Retrieved 7 May 2018.
- Ontanon, Santiago; Synnaeve, Gabriel; Uriarte, Alberto; Richoux, Florian; Churchill, David; Preuss, Mike (December 2013). "A Survey of Real-Time Strategy Game AI Research and Competition in StarCraft". IEEE Transactions on Computational Intelligence and AI in Games. 5 (4): 293–311. doi:10.1109/TCIAIG.2013.2286295.
- "Facebook Quietly Enters StarCraft War for AI Bots, and Loses". WIRED. 2017. Retrieved 7 May 2018.
- Schaeffer, J.; Burch, N.; Bjornsson, Y.; Kishimoto, A.; Muller, M.; Lake, R.; Lu, P.; Sutphen, S. (2007). "Checkers is solved". Science. 317 (5844): 1518–1522. Bibcode:2007Sci...317.1518S. CiteSeerX 10.1.1.95.5393
. doi:10.1126/science.1144079. PMID 17641166.
- "God's Number is 20".
- Bowling, M.; Burch, N.; Johanson, M.; Tammelin, O. (2015). "Heads-up limit hold'em poker is solved". Science. 347 (6218): 145–9. Bibcode:2015Sci...347..145B. doi:10.1126/science.1259433. PMID 25574016.
- "In Major AI Breakthrough, Google System Secretly Beats Top Player at the Ancient Game of Go". WIRED. Retrieved 28 December 2017.
- Sheppard, B. (2002). "World-championship-caliber Scrabble". Artificial Intelligence. 134: 241–275. doi:10.1016/S0004-3702(01)00166-7.
- Tesauro, Gerald (March 1995). "Temporal difference learning and TD-Gammon". Communications of the ACM. 38 (3): 58–68. doi:10.1145/203330.203343.
- Tesauro, Gerald (January 2002). "Programming backgammon using self-teaching neural nets". Artificial Intelligence. 134 (1–2): 181–199. doi:10.1016/S0004-3702(01)00110-2.
...at least two other neural net programs also appear to be capable of superhuman play
- "Kramnik vs Deep Fritz: Computer wins match by 4:2". Chess News. 2006-12-05. Retrieved 2018-07-15.
- "The Week in Chess 771". theweekinchess.com. Retrieved 2018-07-15.
- Nickel, Arno (May 2017). "Zor Winner in an Exciting Photo Finish". www.infinitychess.com. Innovative Solutions. Retrieved 2018-07-17.
... on third place the best centaur ...
- Watson beats Jeopardy grand-champions. https://www.nytimes.com/2011/02/17/science/17jeopardy-watson.html
- Jackson, Joab. "IBM Watson Vanquishes Human Jeopardy Foes". PC World. IDG News. Retrieved 2011-02-17.
- "The Arimaa Challenge". arimaa.com. Retrieved 2018-07-15.
- "AlphaGo beats Ke Jie again to wrap up three-part match". The Verge. Retrieved 2018-07-15.
- Computer bridge#Computers versus humans
- Proverb: The probabilistic cruciverbalist. By Greg A. Keim, Noam Shazeer, Michael L. Littman, Sushant Agarwal, Catherine M. Cheves, Joseph Fitzgerald, Jason Grosland, Fan Jiang, Shannon Pollard, and Karl Weinmeister. 1999. In Proceedings of the Sixteenth National Conference on Artificial Intelligence, 710-717. Menlo Park, Calif.: AAAI Press.
- Wernick, Adam (24 Sep 2014). "'Dr. Fill' vies for crossword solving supremacy, but still comes up short". Public Radio International. Retrieved Dec 27, 2017.
The first year, Dr. Fill came in 141st out of about 600 competitors. It did a little better the second-year; last year it was 65th
- "AI bots trained for 180 years a day to beat humans at Dota 2". The Verge. 25 June 2018. Retrieved 17 July 2018.
- "Waymo's Autonomous Cars Cut Out Human Drivers in Road Tests". The New York Times. 7 November 2017. Retrieved 31 May 2018.
- "On the Road – Waymo". Waymo. Retrieved 2018-07-08.
- Vehicles, California Department of Motor. "Report of Traffic Collision Involving an Autonomous Vehicle (OL 316)". www.dmv.ca.gov. Retrieved 2018-07-08.
- "Microsoft researchers say their newest deep learning system beats humans -- and Google - VentureBeat - Big Data - by Jordan Novet". VentureBeat.
- Santoro, Adam; Bartunov, Sergey; Botvinick, Matthew; Wierstra, Daan; Lillicrap, Timothy (19 May 2016). "One-shot Learning with Memory-Augmented Neural Networks". page 5: Table 1 (Report). arXiv:1605.06065
[cs.LG]. (PDF, 2237kb). Accessed 2017-06-04.
4.2. Omniglot Classification: "The network exhibited high classification accuracy on just the second presentation of a sample from a class within an episode (82.8%), reaching up to 94.9% accuracy by the fifth instance and 98.1% accuracy by the tenth.
- Meyer, Robinson (2015). "How Worried Should We Be About Facial Recognition?". The Atlantic. Retrieved 2 March 2018.
- Antol, Stanislaw, et al. "Vqa: Visual question answering." Proceedings of the IEEE International Conference on Computer Vision. 2015.
- "Artificial Intelligence Index 2017 Annual Report" (PDF). Stanford 100 Year Study on AI. Retrieved 28 December 2017.
- "Robots with legs are getting ready to walk among us". The Verge. Retrieved 28 December 2017.
- Hurst, Nathan. "Why Funny, Falling, Soccer-Playing Robots Matter". Smithsonian. Retrieved 28 December 2017.
- "The Business of Artificial Intelligence". Harvard Business Review. 18 July 2017. Retrieved 28 December 2017.
- Brynjolfsson, E., & Mitchell, T. (2017). What can machine learning do? Workforce implications. Science, 358(6370), 1530-1534.
- There are several ways of evaluating machine translation systems. People competent in a second language frequently outperform machine translation systems but the average person is often less capable. Some machine translation systems are capable of a large number of languages, like google translate, and as a result have a broader competence than most humans. For example, very few humans can translate from Arabic to Polish and French to Swahili and Armenian to Vietnamese. When comparing over several languages machine translation systems will tend to outperform humans.
- Gray, Richard (2018). "How long will it take for your job to be automated?". BBC. Retrieved 31 January 2018.
- "AI will be able to beat us at everything by 2060, say experts". New Scientist. 2018. Retrieved 31 January 2018.
- Grace, K., Salvatier, J., Dafoe, A., Zhang, B., & Evans, O. (2017). When will AI exceed human performance? Evidence from AI experts. arXiv preprint arXiv:1705.08807.
- McClain, Dylan Loeb (11 September 2010). "Bent Larsen, Chess Grandmaster, Dies at 75". The New York Times. Retrieved 31 January 2018.
- "The Business of Artificial Intelligence". Harvard Business Review. 18 July 2017. Retrieved 31 January 2018.
- "4 Crazy Predictions About the Future of Art". Inc.com. 2017. Retrieved 31 January 2018.
- Koch, Christof (2016). "How the Computer Beat the Go Master". Scientific American. Retrieved 31 January 2018.
- "'I'm in shock!' How an AI beat the world's best human at Go". New Scientist. 2016. Retrieved 31 January 2018.
- Moyer, Christopher (2016). "How Google's AlphaGo Beat a Go World Champion". The Atlantic. Retrieved 31 January 2018.
- Johnson, George (29 July 1997). "To Test a Powerful Computer, Play an Ancient Game". The New York Times. Retrieved 31 January 2018.
- Johnson, George (4 April 2016). "To Beat Go Champion, Google's Program Needed a Human Army". The New York Times. Retrieved 31 January 2018.
- "Cracking GO". IEEE Spectrum: Technology, Engineering, and Science News. 2007. Retrieved 31 January 2018.
- "The Mystery of Go, the Ancient Game That Computers Still Can't Win". WIRED. 2014. Retrieved 31 January 2018.
- Gibney, Elizabeth (28 January 2016). "Google AI algorithm masters ancient game of Go". Nature. pp. 445–446. Bibcode:2016Natur.529..445G. doi:10.1038/529445a. Retrieved 31 January 2018.
- Bostrom, Nick (2013). Superintelligence. Oxford: Oxford University Press. ISBN 0199678111.
- Khatchadourian, Raffi (16 November 2015). "The Doomsday Invention". The New Yorker. Retrieved 31 January 2018.
- Müller, V. C., & Bostrom, N. (2016). Future progress in artificial intelligence: A survey of expert opinion. In Fundamental issues of artificial intelligence (pp. 555-572). Springer, Cham.
- Muehlhauser, L., & Salamon, A. (2012). Intelligence explosion: Evidence and import. In Singularity Hypotheses (pp. 15-42). Springer, Berlin, Heidelberg.
- Tierney, John (25 August 2008). "Vernor Vinge's View of the Future - Is Technology That Outthinks Us a Partner or a Master ?". The New York Times. Retrieved 31 January 2018.
- "Superhumanism". WIRED. 1995. Retrieved 31 January 2018.
- "Tech Luminaries Address Singularity". IEEE Spectrum: Technology, Engineering, and Science News. 2008. Retrieved 31 January 2018.
- Molloy, Mark (17 March 2017). "Expert predicts date when 'sexier and funnier' humans will merge with AI machines". The Telegraph. Retrieved 31 January 2018.