Microsoft’s “Bing Predicts” algorithm correctly guessed 21 out of 24 Academy Award winners this year, making it as accurate as the best human experts and more accurate than guesses obtained from polls of thousands of human non-experts.
Wikipedia describes Bing Predicts as “a prediction engine developed by Microsoft that uses machine learning from data on trending social media topics (and sentiment towards those topics), along with trending searches on Bing. It predicts the outcomes of political elections, popular reality shows, and major sporting events.”
Bing Predicts did better than 29 out of 30 professional film critics who posted their 2018 Oscar guesses on the prediction website www.goldderby.com. The highest-performing human expert, sports host and film critic Adnan Virk, tied Bing Predicts by correctly guessing 21 of the Award winners.
Moreover, Bing Predicts’ accuracy rate of 87.5% (21/24) surpassed the 85.7% accuracy rate (6/7) derived from a large survey of human non-experts. New Scientist magazine conducted the survey to test the value of the “wisdom of the crowd” for predicting future events, and over 6,000 people participated.
Bing Predicts’ accuracy has been high for previous Academy Awards, though this year was one of its best:
- 2017 Oscars: 17/24
- 2016 Oscars: 17/24
- 2015 Oscars: 20/24
- 2014 Oscars: 21/24
Bing also excels at predicting outcomes of elections, sports games, and other major competitions. A few examples include:
- Bing Predicts made nearly perfect guesses about the outcomes of soccer matches at the 2014 World Cup.
- Bing Predicts correctly guessed Scotland would vote to stay in the U.K. in the 2014 referendum
- Bing Predicts guessed the outcomes of the U.S. 2014 Midterm elections with 95% accuracy.
I have to wonder, if machines continue improving their powers of prediction and their intelligence, will they someday suck the fun, risk and mystery out of every aspect of life?
For example, what would be the fun in watching sports if the outcome of every event were known with 99% certainty beforehand? How would our lives change if AIs constantly calculated and told us of the odds of success for every action–big (career choice or marriage) or small (where to eat lunch)–that we were contemplating? Taken to an even greater extreme, what would it be like if machines intervened to prevent us from making self-destructive or even just sub-optimal choices, and always impelled us towards the safe course of action?
As constraining and un-free as such a future might sound, what if it were demonstrably superior in terms of allocating human labor, and achieving some optimal balance between maximized productivity, maximized average happiness, and minimization of extreme human suffering? It would certainly be in keeping with the long-running historical trend for things overall to improve with time, while narrower aspects of life (such as overcrowding or certain types of pollution) worsen.
Links
- https://www.msn.com/en-us/movies/oscars/bing-predicts-who-will-win-the-oscars/ss-BBJvArd
- http://www.goldderby.com/article/2018/oscars-2018-adnan-virk-espn-tops-all-experts-predicting-winners/
- https://www.newscientist.com/article/2162869-how-new-scientist-readers-predicted-the-oscar-winners/
- http://www.thedrum.com/news/2017/02/24/and-the-oscar-goes-la-la-land-if-bing-predicts-right
- http://www.businessinsider.com/microsoft-bing-oscars-2016-2016-2
- https://www.theverge.com/2015/2/23/8089593/microsoft-bing-oscar-predictions
- https://www.geekwire.com/2014/microsoft-bing-15-16-world-cup/
- https://qz.com/267900/scotland-independence-referendum-prediction/
- https://www.networkworld.com/article/2846248/microsoft-subnet/forget-the-pollsters-microsofts-bing-predicted-midterm-election-with-95-accuracy.html