Will Humanity deploy any new technology in an unsafe way by 2030? (Contra LeCun)
23
Ṁ1259
2030
92%
chance

Context:

Resolves Yes if any counterexample to Yann's claim here occurs before 2030, in my judgement. Resolves No otherwise.

I will not trade in this market.

Get Ṁ1,000 play money
Sort by:

But can anyone convince Lecun of this fact?

predicts YES

Yud is a rascal and I feel like this is an appropriate addition to this market's commentary?

https://twitter.com/ESYudkowsky/status/1739776629029876213

Aren’t there trivially thousands of examples of technologies being used unsafely? Drugs? Weapons? Pesticides? Hacking? Cars?! don’t get it.

Minor quibble but a positive resolution of your question doesn’t necessarily refute LeCun since he conditionalizes on the existence of a safe pathway. We could deploy a new technology in an unsafe way because there is no safe way.

@NicoDelon Good points all around. I'm open to refining the resolution criteria here, but I'm not trying to "gotcha" LeCunn. That said, yes this should still be trivial I think? But note it has to be a "new technology" which is itself fuzzy.

Going through your list to try to think of what would/wouldn't count:

Drugs - The Opioid epidemic would be a counterexample I think. Same with cigarettes as NC pointed out. But any new opioids after this market are probably not a "new technology" unless they're like some super opioid that claims they're non-addictive, and then it turns out they are addictive.

Weapons - This is a bit of a gotcha, I'd probably only count a weapon as a new technology deployed unsafely if it was like a newsmaking new weapon that exploded and blew your hands off. It's supposed to be unsafe to the people you point it at.

Pesticides - Also probably counts a counterexample in the past? I don't actually know that much about pesticides and how harmful they are, or how harmful they were when first deployed. A new fancy kind of pesticide that claims to be safe and then turns out to make the produce poisonous could count.

Hacking - Not sure what you mean on this one.

Cars - Cars are very much an old technology and known to have many unsafe features, so I think unlikely to count as a new counterexample. But maybe if Cybertrucks started routinely blowing up when you try to roll down the window or something like that.

And yeah, in all these cases I'm listing I think there would be a safe way to have done things and so it's a counterexample because it's done unsafely.

bought Ṁ1 YES at 93%
predicts YES

@Joshua Sorry, I didn’t mean to use those as examples of new technologies but as a way to pin holes in the implicit base he seems to be using for induction. I think we agree it is very likely that any new technology has the potential to be misused and that, if it can, it likely will. I find his confidence quite alarming (and I’m not a Luddite or a fan of precautionary principles).

Will this be judged at the end of 2030? Seems to me that a lot of harmful technologies aren't detected as harmful until decades down the line (e.g. cigarettes, leaded petrol).

@NcyRocks

I'd resolve it early if there's a clear case, but could wait to resolve if there is an ambiguous case like those.

@Joshua I don't think they were ambiguous when they were first introduced.

everyone's looking pretty stressed in that dream image

Tag yourself I'm


I'll be the well-dressed three-armed baby pondering whether Yud was right or if I'm born just in time for a transhuman utopia
(coinflip market)