2. This is the article: 3. An AI is only as powerful as it’s ability to act in t

2. This is the article: https://www.sciencemag.org/news/2020/11/game-has-changed-ai-triumphs-solving-protein-structures

3. An AI is only as powerful as it’s ability to act in the real world.

4. It cannot act in the real world without a model of the real world.

5. Current AI has no real world model, and so can’t LEARN w/o massive precompiled data.


Source date (UTC): 2020-11-30 19:00:32 UTC

Original post: https://twitter.com/i/web/status/1333486085108797446

Replying to: https://twitter.com/i/web/status/1333486081896050689


IN REPLY TO:

Unknown author

1. I want to use this article on AI solving the protein problem (what I consider closest to physics problem) to illustrate where AI is useful, and why it’s weak: because hard problems are rarely computational. Instead hard problems require a lot of experiments before computation.

Original post: https://x.com/i/web/status/1333486081896050689

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *