- Thanks to leaps and bounds in the field of artificial intelligence in the past decade, robots are increasingly beating humans at our own games.
- AI-powered programs have proven their prowess at competitive games and academic tests alike throughout the past 10 years.
- Many advances in AI can’t be quantified with competitions or challenges, but robots’ victories at games ranging from Jeopardy to Dota show how far AI has come.
- Visit Business Insider’s homepage for more stories.
When IBM’s Deep Blue chess machine defeated world chess champion Garry Kasparov in 1997, the world responded with surprise and angst at how far computers had come: “Be Afraid,” read a Weekly Standard headline reacting to the news.
Artificial intelligence has since made advancements that were unthinkable just 20 years ago – in the past decade alone, robots have achieved dominance over humans in games far more complex than chess.
While most of those advances can’t be quantified with milestones like chess victories, programmers have continued the tradition of building machines designed to outsmart humans at our own games.
Here’s a comprehensive list of the competitions, games, and challenges that robots beat humans at in the past decade.
2011: IBM’s Watson beats two former champions to win Jeopardy
Watson defeated 74-time Jeopardy winner Ken Jennings and 20-time winner Brad Rutter after a three-day contest, showcasing the strength of IBM’s supercomputer.
2014: Facebook’s DeepFace facial recognition algorithm achieves an accuracy rate of 97%, rivaling the rate of humans
Facial recognition technology has only become more sophisticated since Facebook achieved that milestone in 2014. However, some research has questioned whether human facial recognition can accurately be compared to AI facial recognition.
2015: Google DeepMind’s AlphaGo defeats Go champions in Korea and Europe
In the years following AlphaGo’s 2015 victories, it went on to defeat several other international champions, and by 2017 it was able to win 60 rounds of Go back-to-back.
2016: Microsoft speech recognition AI can transcribe audio with fewer mistakes than humans
Microsoft said that its software achieved an error rate of 0.4%, compared to the human error rate of 5.9%.
2017: Libratus, an AI bot, defeats four of the world’s leading poker players in a 20-day tournament
Unlike Chess or Go, poker is an imperfect information game, meaning players have to guess each others’ hands – making Libratus’ victory all the more impressive.
2017: An OpenAI bot defeats a human esports player at the multiplayer online battle arena game Dota 2
The bot, developed by OpenAI, has repeatedly beaten the world’s top Dota 2 players.
2017: An AI system developed by researchers at Northwestern is able to beat 75 per cent of Americans at a visual intelligence test
The program can solve logic games simply by looking at them – while most vision-based AI at the time focused on recognition, this one takes an extra step with visual reasoning.
2018: Alibaba’s AI outscores humans in a Stanford University reading comprehension test
After answering 10,000 reading comprehension questions, Alibaba’s AI scored 82.44, just above the 82.304 score achieved by humans.
Business Insider Emails & Alerts
Site highlights each day to your inbox.