Colby Cosh: The lifelike nature of artificial intelligence

Interesting test:

…Well, fast-forward a dozen centuries, and along come Copernicus asking “What if Earth isn’t at the centre after all?”; Kepler asking “What if the orbits aren’t circular, but elliptical?”; and Newton, who got to the bottom of the whole thing by introducing the higher-level abstraction of gravitational force. Bye-bye epicycles.

None of these intellectual steps, mind you, added anything to anyone’s practical ability to predict planetary motions. Copernicus’s model took generations to be accepted for this reason (along with the theological/metaphysical objections to the Earth not being at the centre of the universe): it wasn’t ostensibly as sophisticated or as powerful as the old reliable geocentric model. But you can’t get to Newton, who found that the planets and earthbound objects are governed by the same elegant and universal laws of motion, without Copernicus and Kepler.

Which, in 2025, raises the question: could a computer do what Newton did? Vafa’s research group fed orbital data to AIs and found that they could correctly behave like ancient astronomers: make dependable extrapolations about the future movements of real planets, including the Earth. This raises the question whether the algorithms in question generate their successful orbital forecasts by somehow inferring the existence of Newtonian force-abstractions. We know that “false,” overfitted models and heuristics can work for practical purposes, but we would like AIs to be automated Newtons if we are going to live with them. We would like AIs to discover new laws and scientific principles of very high generality and robustness that we filthy meatbags haven’t noticed yet.

When Vafa and his colleagues found is that the AIs remain in a comically pre-Copernican state. They can be trained to make accurate predictions by being presented with observational data, but it seems that they may do so on the basis of “wrong” implicit models, ones that depend on mystifying trigonometric clutter instead of the beautiful inverse-square force law that Newton gave us. The epicycles are back!

The paper goes on to do more wombat-dissecting, using the game of Othello to show how AI reasoning can produce impressive results from (apparently) incomplete or broken underlying models. It is all very unlike the clean, rigorous “computing science” of the past 100 years: whatever you think of the prospects of AI, it is clear that the complexity of what we can create from code, or just buy off the shelf, is now approaching the complexity of biological life.

Source: Colby Cosh: The lifelike nature of artificial intelligence