these "folks" were wrong in advance, none of these things was seriously considered a huge breakthrough. Let me explain...https://twitter.com/Miles_Brundage/status/1399448341671387138 …
Show this thread
There is an entire genre of people (with a vested interested in hyping up AI progress) saying "AI achieved X which experts said would never happen", when, if you look closely, no actual expert was saying X would never happen and most experts were expecting it to happen imminently
-
-
Here's one example of something I've been eyeing since ~2009 as a potential Big Deal for AI (not that I was an expert, nor am I one now): RoboCup. https://en.wikipedia.org/wiki/RoboCup Team of autonomous robots winning the World Cup. Features embodiment, etc.
-
The prediction of those who started RoboCup (experts?), at the time, was that it would be possible by ~2050. Still 30 years to go. I expect it will likely take a bit longer, though 2050 isn't too unrealistic.
- Show replies
New conversation -
-
-
Many Go experts said at the time that AI solved the game of Go 10 years too early. More recently, many protein folding experts have said that Alphafold has achieved what they thought would be possible in a few decades.
-
There are clearly areas where AI is achieving feats that were thought impossible in the short term. Maybe we should create a site that lists the statements of experts in the field where AI impresses, so there will be no more debate.
End of conversation
New conversation -
-
-
I mean I could make it happen too if I could spend 2 billion on the problem like certain companies :) nobody sane in their mind would spend all those money on Go or playing games.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I don't know. Go was quite of dicy, in most expert opinion when it happened. Same with StarCraft. There were 2015 survey of machine learning experts and very few expected those to happen before 2020. The timelines for champion level go were estimated for between 2020-2040.
-
I was genuinely very surprised that StarCraft was solved under these (APM, vision) constraints. It was a ridiculous accomplishment.
End of conversation
New conversation -
-
-
What do experts say right now about lure vision level 5 autonomy?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
A variation of this that I often see is when experts in other fields (Physics, Philosophy, Maths, etc) believe certain problems are "AI Complete" (requiring a general AI to solve them) and then the problems are promptly solved without it. Nick Bostrom comes to mind.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.