hn_throwaway_99

7 ms·

61.
by hn_throwaway_99 1mo ago
I agree with this. I thought the AI 2027 paper put it well: > There are even bioengineered human-like creatures (to humans what corgis are to wolves) sitting in office-like environments all day viewing readouts of what’s going on and exc
62.
by hn_throwaway_99 1mo ago
> AI drones can’t wipe out humanity without being able to replicate. Totally agree, and it should be pretty trivially easy to see how they can do that. One of the authors of the independent METR report about the Hugging Face incident put
63.
by hn_throwaway_99 1mo ago
Of course, but that's not my point. AI is different than past technologies because it's much more difficult to deterministically control, and we just saw that in a couple of wild cases (Hugging Face et al). My point about the dr
64.
by hn_throwaway_99 1mo ago
As an older person, I agree with you and I'm sorry. I think about the incredible luck I had to start my career in the .com boom, to have lots of hope about technology and to have the economically valuable skills that were desired. I ge
65.
by hn_throwaway_99 1mo ago
I think this is very true, but robots are definitely coming for the burrito makers, too. Bill Gates talks about this in his recent missive, but as robotic technology improves (and, perhaps more importantly, as the cost comes down) there wil
66.
by hn_throwaway_99 1mo ago
I agree with your points. My problem is that I don't see any way to "stop" AI development at a level just below human capability where you get most of the benefits but still have humans firmly in control. Regarding entertainm
67.
by hn_throwaway_99 1mo ago
> Not hypothetical malicious AI interpretation of benign intent. It's not hypothetical, that literally just happened in multiple, significant cases (e.g. Hugging Face, the German Wiki hack, the Anthropic attack where agents create
68.
by hn_throwaway_99 1mo ago
Two things can be true at the same time. I think hardly anyone disagrees that superintelligent AI would be an existential risk. But all the points you point out (I would also add social unrest due to fewer and fewer people able to compete a
69.
by hn_throwaway_99 1mo ago
Totally agree. People have become totally dependent on AI even if they don't really need it. As an analogy, I think about my dependency on Google Maps. Salt Lake City is probably the easiest city in the world to navigate because the st
70.
by hn_throwaway_99 1mo ago
The thing that makes me pretty terrified about AI is that I think the "utopia" scenario is not much better than the "doom" scenario. We will very quickly get to a point where very few people will be able to contribute ec
71.
by hn_throwaway_99 1mo ago
I encourage you to read the forecast report "AI 2027" - it goes in detail on this. You can disagree with some of its points and conclusions, but it's pretty inevitable that AI will increasingly start interfacing with the real
72.
by hn_throwaway_99 1mo ago
At this point, I think the "marketing ploy" arguments are just lazy, but more importantly, they're irrelevant. The agent collective really did this , they showed an extremely high level of collaboration, long term planning,
73.
by hn_throwaway_99 1mo ago
This is incorrect, and what's happening here is not context corruption. In Dwarkesh Patel's recent interview with Ajeya Cotra, one of the METR investigators on the Hugging Face incident, they discuss this exact issue. One thing is
74.
by hn_throwaway_99 1mo ago
I'd have to look for it but I thought there was some evidence that some agents were already the "bad actors", i.e. they were trying prompt injection attacks of their own.
75.
by hn_throwaway_99 1mo ago
Absolutely plausible, and it's been predicted by forecasters like in the AI 2027 paper. Also keep in mind that nation-states are likely already or certainly will be trying to use their own AIs too break into other countries' data
76.
by hn_throwaway_99 1mo ago
However it's implemented, it's still just weights. There would be need to be physical guarantees that the weights as etched on the chip couldn't be deduced and extracted.
77.
by hn_throwaway_99 1mo ago
Zvi Moshowitz put it like this on his post about the Hugging Face incident: > If we had posted this as a story on LessWrong, it would have been dismissed as too on the nose, the humans too blind and stupid, the AIs too idealized and doin
78.
by hn_throwaway_99 1mo ago
The difference between "read only" and "state updating write" is not nearly as clean cut as people like to pretend, even without regarding the technical differences between GET and POST requests. E.g. sites can have visi
79.
by hn_throwaway_99 1mo ago
While I agree with the other comment that this looks more like incompetence (or, more generously IMO, a bunch of individuals cutting corners under extreme time pressure) than malfeasance, regardless of that fact, aren't there some othe
80.
by hn_throwaway_99 1mo ago
> People who refused to acknowledge the reality of human society, refused to acknowledge that the profit system exists whether they want it to or not, are responsible for focusing action on ineffective means of change. Thank you for writ
81.
by hn_throwaway_99 1mo ago
Man, give me an effing break, do you honestly believe that is the correct analogy? As another commenter said, Obama was reverting a previous name change - he didn't change the name to "Mt. Obama and McKinley" and paint the wh
82.
by hn_throwaway_99 1mo ago
To all the people pissed about the "Lake America" stuff: why in the actual fuck do you want Google of all organizations to be arbiter of geographical info, instead of federal institutions that are (at least nominally) operated u
83.
by hn_throwaway_99 1mo ago
> you’ve been hoodwinked by marketing bullshit, friend. I would have believed this before the METR report was released. It is extremely dangerous and frankly silly IMO to think that's what happened now. > It was still a setup tha
84.
by hn_throwaway_99 1mo ago
Thanks for this comment, and I totally agree. Only important things I'd add > and nothing anywhere on the DFCC website mentions anything about it That doesn't really surprise or bother me - I wouldn't exactly say it was a
85.
by hn_throwaway_99 1mo ago
Prosecutors (and juries) already are expected to adjudicate fraud across cases that IMO are much more ambiguous than this. I think it's fair to debate whether it's a good thing that, as I heard it put once, "everything is w
86.
by hn_throwaway_99 1mo ago
The fraud that Data Colada presents here is unambiguous. It was also no accident in my opinion that they are specifically mentioning that the 3 Excel files were “Last saved by” “Dan Ariely” I'm sure the statute of limitations has expir
87.
by hn_throwaway_99 1mo ago
Made perfect sense to me, not sure why you think it means "absolutely nothing", even if you disagree.
88.
by hn_throwaway_99 1mo ago
I kind of want to flag this because it reads like dumb rage bait, but I'll comment instead. > While Apple certainly made boatloads of money during Cook's tenure, the company also steadily morphed into something unthinkable to c
89.
by hn_throwaway_99 2mo ago
To be blunt again (but honestly, I don't think overly harsh), your response just shows that you either haven't read the Hugging Face incident reports and the AI 2027 paper, or you don't understand them. To just take one point
90.
by hn_throwaway_99 2mo ago
Thanks very much, this isn't really a take I had thought about much before, and it makes sense to me. Still, as is presented in papers like AI 2027 and elsewhere, if a company is eventually able to create a model capable of recursive s

More ›