Макс Тегмарк - Life 3.0 - Being Human in the Age of Artificial Intelligence

We’re the guardians of the future of life now as we shape the age of AI. Although I cried in London, I now feel that there’s nothing inevitable about this future, and I know that it’s much easier to make a difference than I thought. Our future isn’t written in stone and just waiting to happen to us—it’s ours to create. Let’s create an inspiring one together!

* This experience also made me rethink how I personally should interpret news. Although I’d obviously been aware that most outlets have their own political agenda, I now realized that they also have a bias away from the center on all issues, even nonpolitical ones.

Notes

Chapter 1

1. “The AI Revolution: Our Immortality or Extinction?” Wait But Why (January 27, 2015), at http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html.

2. This open letter, “Research Priorities for Robust and Beneficial Artificial Intelligence,” can be found at http://futureoflife.org/ai-open-letter/.

3. Example of classic robot alarmism in the media: Ellie Zolfagharifard, “Artificial Intelligence ‘Could Be the Worst Thing to Happen to Humanity,’ ” Daily Mail, May 2, 2014; http://tinyurl.com/hawkingbots.

Chapter 2

1. Notes on the origin of the term AGI: http://wp.goertzel.org/who-coined-the-term-agi.

2. Hans Moravec, “When Will Computer Hardware Match the Human Brain?” Journal of Evolution and Technology (1998), vol. 1.

3. In the figure showing computing power versus year, the pre-2011 data is from Ray Kurzweil’s book How to Create a Mind, and subsequent data is computed from the references in https://en.wikipedia.org/wiki/FLOPS.

4. Quantum computing pioneer David Deutsch describes how he views quantum computation as evidence of parallel universes in his The Fabric of Reality: The Science of Parallel Universes—and Its Implications (London: Allen Lane, 1997). If you want my own take on quantum parallel universes as the third of four multiverse levels, you’ll find it in my previous book: Max Tegmark, Our Mathematical Universe: My Quest for the Ultimate Nature of Reality (New York: Knopf, 2014).

Chapter 3

1. Watch “Google DeepMind’s Deep Q-learning Playing Atari Breakout” on YouTube at https://tinyurl.com/atariai.

2. See Volodymyr Mnih et al., “Human-Level Control Through Deep Reinforcement Learning,” Nature 518 (February 26, 2015): 529–533, available online at http://tinyurl.com/ataripaper.

3. Here’s a video of the Big Dog robot in action: https://www.youtube.com/watch?v=W1czBcnX1Ww.

4. For reactions to the sensationally creative line 5 move by AlphaGO, see “Move 37!! Lee Sedol vs AlphaGo Match 2,” at https://www.youtube.com/watch?v=JNrXgpSEEIE.

5. Demis Hassabis describing reactions to AlphaGo from human Go players: https://www.youtube.com/watch?v=otJKzpNWZT4.

6. For recent improvements in machine translation, see Gideon Lewis-Kraus, “The Great A.I. Awakening,” New York Times Magazine (December 14, 2016), available online at http://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html. GoogleTranslate is available here at https://translate.google.com.

7. Winograd Schema Challenge competition: http://tinyurl.com/winogradchallenge.

8. Ariane 5 explosion video: https://www.youtube.com/watch?v=qnHn8W1Em6E.

9. Ariane 5 Flight 501 Failure report by the inquiry board: http://tinyurl.com/arianeflop.

10. NASA’s Mars Climate Orbiter Mishap Investigation Board Phase I report: http://tinyurl.com/marsflop.

11. The most detailed and consistent account of what caused the Mariner 1 Venus mission failure was incorrect hand-transcription of a single mathematical symbol (a missing overbar): http://tinyurl.com/marinerflop.

12. A detailed description of the failure of the Soviet Phobos 1 Mars mission can be found in Wesley T. Huntress Jr. and Mikhail Ya. Marov, Soviet Robots in the Solar System (New York: Praxis Publishing, 2011), p. 308.

13. How unverified software cost Knight Capital $440 million in 45 minutes: http://tinyurl.com/knightflop1 and http://tinyurl.com/knightflop2.

14. U.S. government report on the Wall Street “flash crash”: “Findings Regarding the Market Events of May 6, 2010” (September 30, 2010), at http://tinyurl.com/flashcrashreport.

15. 3-D printing of buildings (https://www.youtube.com/watch?v=SObzNdyRTBs), micromechanical devices (http://tinyurl.com/tinyprinter) and many things in between (https://www.youtube.com/watch?v=xVU4FLrsPXs).

16. Global map of community-based fab labs: https://www.fablabs.io/labs/map.

17. News article about Robert Williams being killed by an industrial robot: http://tinyurl.com/williamsaccident.

18. News article about Kenji Urada being killed by an industrial robot: http://tinyurl.com/uradaaccident.

19. News article about Volkswagen worker being killed by an industrial robot: http://tinyurl.com/baunatalaccident.

20. U.S. government report on worker fatalities: https://www.osha.gov/dep/fatcat/dep_fatcat.html.

21. Car accident fatality statistics: http://tinyurl.com/roadsafety2 and http://tinyurl.com/roadsafety3.

22. On the first Tesla autopilot fatality, see Andrew Buncombe, “Tesla Crash: Driver Who Died While on Autopilot Mode ‘Was Watching Harry Potter,’ ” Independent (July 1, 2016), http://tinyurl.com/teslacrashstory. For the report of the Office of Defects Investigation of the U.S. National Highway Traffic Safety Administration, see http://tinyurl.com/teslacrashreport.

23. For more about the Herald of Free Enterprise disaster, see R. B. Whittingham, The Blame Machine: Why Human Error Causes Accidents (Oxford, UK: Elsevier, 2004).

24. Documentary about the Air France 447 crash: https://www.youtube.com/watch?v=dpPkp8OGQFI; accident report: http://tinyurl.com/af447report; outside analysis: http://tinyurl.com/thomsonarticle.

25. Official report on the 2003 U.S.-Canadian blackout: http://tinyurl.com/uscanadablackout.

26. Final report of the President’s Commission on the Accident at Three Mile Island: http://www.threemileisland.org/downloads/188.pdf.

27. Dutch study showing how AI can rival human radiologists at MRI-based diagnosis of prostate cancer: http://tinyurl.com/prostate-ai.

28. Stanford study showing how AI can best human pathologists at lung cancer diagnosis: http://tinyurl.com/lungcancer-ai.

29. Investigation of the Therac-25 radiation therapy accidents: http://tinyurl.com/theracfailure.

30. Report on lethal radiation overdoses in Panama caused by confusing user interface: http://tinyurl.com/cobalt60accident.

31. Study of adverse events in robotic surgery: https://arxiv.org/abs/1507.03518.

32. Article on number of deaths from bad hospital care: http://tinyurl.com/medaccidents.

33. Yahoo set a new standard for “big hack” when announcing a billion(!) of their user accounts had been breached: https://www.wired.com/2016/12/yahoo-hack-billion-users/.

34. New York Times article on acquittal and later conviction of KKK murderer: http://tinyurl.com/kkkacquittal.

35. The Danziger et al. 2011 study (http://www.pnas.org/content/108/17/6889.full), arguing that hungry judges are harsher, was criticized as flawed by Keren Weinshall-Margela and John Shapard (http://www.pnas.org/content/108/42/E833.full), but Danziger et al. insist that their claims remain valid (http://www.pnas.org/content/108/42/E834.full).

36. Pro Publica report on racial bias in recidivism-prediction software: http://tinyurl.com/robojudge.

37. Use of fMRI and other brain-scanning techniques as evidence in trials is highly controversial, as is the reliability of such techniques, although many teams claim accuracies better than 90%: http://journal.frontiersin.org/article/10.3389/fpsyg.2015.00709/full.