Evidence receipt / belief
Published · transcript-backedGary Marcus: belief
24 Jun 2025 Machine Learning Street Talk Three Red Lines We're About to Cross Toward AGI (Daniel Kokotajlo, Gary Marcus, Dan Hendrycks)
“I hadn't played this variation either before, but it was obvious what to do. So it failed to identify what 3 in a row was, which I would say goes to a conceptual weakness like you should have had enough data to understand what 3 in a row was.”
Source trail
Everything needed to verify it.
- Speaker
- Gary Marcus
- Attribution
- Verified speaker
- Claim type
- belief
- Recorded
- 24 Jun 2025
- Publisher
- Machine Learning Street Talk
Transcript context
…So I haven't heard about this. I had a friend look into this and he's very easily able to do it. I can give you some other example. Let me just put 1 other example on the table, a similar flavor, which is I played grok in Tic Tac Toe and I gave came up with some variation. And then, I might write this up this weekend. And so I asked some variation. We played it and then it offered another variation, which was let's only play tic tac toe where you can let's play tic tac toe where you can only win if your 3 in a row is on the edges. And I said fine. We had some moves back and forth. And then it suggested some moves for me at a point where I could in fact make a 3 in a row. So it had failed to develop the right strategy. I hadn't played this variation either before, but it was obvious what to do. So it failed to identify what 3 in a row was, which I would say goes to a conceptual weakness like you should have had enough data to understand what 3 in a row was. And so he it suggested other moves didn't recognize it. It's a pretty bad failure. And then we played again after I corrected it. You know how you get these sycophantic answers, I'm sorry, you know. So so we played again and it lost to me again. And then a third time. I posted all of this on Twitter a couple weeks ago. So if you move these things off the typical problem, they're even worse. If they had a robust understanding of a domain like tic tac toe or chess, it wouldn't be too hard. I posted another on Twitter, which was like, I had some variation on chess, like, give me a board state where there are 3 queens on white side, but black, can immediately win. And performance was not great. So like you've you take knowledge that for an expert would be out of what they've specifically practiced, but they understand the domain well enough. And experts will have no trouble at all. And these systems still do. So I totally agree with those limitations about current systems. But the thing that I would say is that it seems like the trends are going up. So I would predict that if you measured this sort of thing for the last 5 years, even though the current systems might still have weaknesses, they would be less weak than the systems of 2 years ago, which are…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.