What are some tasks where AI exceeded human expectations?

There have been many tasks that people in the past thought would be very hard or impossible for AI, which AI can now do.

Chess

Edgar Allan Poe wrote about The Turk and posited that no pure machine could adapt to the unpredictable complexities of chess. He was correct that The Turk was not a pure machine, but eventually his thesis that no machine could play chess was proven wrong.

Go

When chess champions were defeated by Deep Blue in 1997, it was suggested that the game of Go would resist computers for some time due to its much larger number of possible positions. 20 years later, Deepmind’s AlphaGo (and later the more general AlphaZero) surpassed humans. It’s worth noting that even such superhuman players are still vulnerable to adversarial attacks.

CAPTCHAs

CAPTCHAs are visual tests that are meant to be hard for computers but easy for humans to complete. These tests have repeatedly needed to be redesigned as computer algorithms improved, and LLMs seem to be able to solve many of them.

Art

Some have thought that computers would never be able to create art. Recent uses of diffusion modelsn have allowed the generation of visual art (Midjourney, DALL-E and Stable Diffusion generate images, Sora generates video), and music (suno.ai) that is sometimes indistinguishable from the output of a skilled human.

Turing tests

For some time, the Turing Test (which tests whether a human observer can distinguish a machine from a human based on textual outputs from each) was considered the ultimate test in differentiating humans from computers, but it has been argued that LLMs quietly passed them, and that we need new forms of this test. It’s worth noting that humans have been fooled by computers pretending to be humans for over 50 years.