I think what we're seeing is an explosion of new approaches to computerized problem solving made possible by huge amounts of data and enormous computing resourc

I think what we're seeing is an explosion of new approaches to computerized problem solving made possible by huge amounts of data and enormous computing resources. A lot of what has become possible in the last couple of decades is indeed new, but the apparent rapid advance is really just a matter of applying the brute force of a massively upgraded ability to process huge quantities of data in parallel, and this has led us to make erroneous assumptions about future progress in these areas. Basically, these are new solutions to new problems, and we're rapidly seeing the easy 80% of this new generation of "AI" happen and it seems magical. But soon enough we'll hit the wall where further progress becomes harder and harder and brute force approaches are no longer sufficient to achieve interesting results.