"even safer" in this case means some combination of two things:
The new organization is more ideologically aligned with the transhumanist doom cult that apparently managed to eat the brains of the people with money to burn.
The new organization, largely as a result of this, is capable of sinking an unending amount of capital into buying compute time and Nvidia chips but due to their commitments to safety is even less inclined to actually deliver anything.
I found the comment about models creating very old-fashioned "18th century style" proofs very interesting. Not surprising in retrospect since older proofs are going to be reproduced more across the training data compared to newer ones, but it's still interesting to note and indicative of the reproduction that these things are doing.
I would go so far as to try and find a suitably precocious undergrad to run the test that they themselves are capable of guiding and nudging the model the way OpenAI's team did but not of determining on their own that the conjecture in question is false. OpenAI's results here needed a fair bit of cajoling and guidance, and without that I can only assume it would give the same kind of non-answer regardless of whether the question is in fact solvable.
The point about heavy artillery is actually pretty salient, though a more thorough examination would also note that "Lethal Autonomous Weapons Systems" is a category that includes goddamn land mines. Of course this would serve to ground the discussion in reality and is thus far less interesting to people who start organizations like the Future of Life Institute.
I wonder if this is going to hold out long enough to get some obnoxious AI-first language created that is designed to have as obnoxiously picky of a compiler as it can in order to try and turn runtime errors that the model can't cope with into compile failures which it can silently retry until they're 'fixed'
Today in excellent cold opens: "I didn't talk to ChatGPT, I never have. Instead, I took a load of edibles and laid down in the driveway with the hose on. I produced nothing of value and wasted a ton of water, but at least I ate three protein bars so I'm so healthy."
Between this shit and the oncoming tech-inflicted recession, my wife and I are both reskilling out of tech. I'm looking into electrician and she's looking into accounting. Two fields that are, at least in theory, sufficiently motivated by accuracy and reality to be at least somewhat protected from the rise and fall of the confabulatron.
We'll see how it goes, but God if that "Burn in hell" doesn't just hit me right in the soul.
I had to try and talk my wife back from the edge a little bit the other night and explain the difference between reading the published evidence of an actual conspiracy and qanon-style baking. It's so easy to try and turn Epstein into Evil George Soros, especially when the real details we have are truly disturbing.
Now I'm just imagining an AI quarterback and the whole team revolting at following plays called by something that won't end up at the bottom of the 1000lb pile of meat if they fuck it up.
Also despite being a less-than-zero effort attack please note that as of sharing we have one successful "corruption" (i.e. a comment about zucchini and API keys) and two comments from bots too stupid to coherently understand the OP at all.
The AI bubble keeps radicalizing me further by letting me know that there exists a class of people who are able and willing to pay $300 a day to have a stupid chatbot expose their PII to hackers and maybe set appointments for them just so long as they don't have to interact with a poor person for whom that would be a life-changing amount of money.
The accountability sink backfires since in order to sell this they have to acknowledge how much of a role they play in the process, rather than being simply a platform that connects applicants to openings and doesn't make decisions.
It makes me think that they're sufficiently poorly designed that it's treating the reset as a temporary communication issue. I wonder if you could use this to their detriment by configurating the server to silently drop the connection rather than RSTing it. From your server's side it should look fairly similar, but from their side they actually have to spend the time putting together and sending the HTTP request before getting shut down.
Yeah. There's something altogether disgusting about people looking at the sheer amount of resources and infrastructure that we as a collective society are pouring into this crap and lamenting that not enough people use them as goddamn toys, even though those are also the only people who don't seem to hate every interaction.
If you take the raw words and ignore the context it's not too much worse, save the first person narrator, but with the context of the author it's goddamn horrifying.
"even safer" in this case means some combination of two things: