Finally.
lol. Desperate marketing.
Who do you suppose they are marketing to? People who want to die? People who want to see an extinction event?
People with a boner about how powerful it is
This is basically marketing for it. With the quiet part being “we will absolutely make sure it stays powerful, but make sure it doesn’t kill us all”
Legislatures who can put them ahead in the market
No such thing as bad press.
Anyone and everyone, though firstly a credulous media… in order to contribute to the huge scam: that this is a product which has some value commensurate with the spending, let alone has any utility. Falsely claiming this faulty overbloated software product is intelligent is their sales proposition in the first place, so messaging that its ‘going rogue’ , acting of its own accord, and other such bs contributes to that. The lack of decent questioning journalism , rather than parrotting pr statements, makes this very cheap marketing but almost almost at zeitgeist level.
People who want to believe Anthropics AI is powerful.
The more senior the employee, the more concerned they are
Peter principal lol
Humans could cause extinction of AI companies and their investors.
And it’s always “could”. Not that it will, nothing specific, just bubble blowing bluffing and bamboozling.
I feel like their contributions to climate change make this true.
I agree, but not on the timescale they mentioned
Hyperbole, but still, I wonder what sort of world I’m bringing my child up into.
oh, …better invest now!
Hey, you absolute fuckers: if this is even remotely true, why the fuck are you not sabotaging this shit? You, as the researchers? Poison data, manipulate algorithms, fucking bomb the data centers, but do something.
Or just put useable guardrails on it.
I have to use Claude for work, if they were actually worried about safety it would be far easier to control the permissions of processes instead they basically let you filter strings that it executed in a subshell.
Modern computing systems can be pretty fine grained around network and systemcall access (including inheritance to prevent escapes), none of these tools do that, because this is just marketing not real fear.
There’s been multiple pretty interesting exposés on this. The TL;DR is that a lot of the high profile figures in the AI space have a god complex where they believe AI is inevitable and by being involved in the development, they might be able to steer it in a better direction. So really, the classic “if I don’t, someone else will” excuse.
Ah yes, Roko’s Basilisk.
They’re not doing anything about it for the same reason you’re not doing anything about it.
I don’t have the same access as they do
Not soon enough.
I wonder if that just means that many of them also don’t understand LLMs?
In general, the more senior the employee, the more concerned they are.
Maybe they feel some underlying stress because subconsciously they know there is a bubble that’s going to burst? And more senior what - business people or people who understand those systems? If the latter: why don’t they give concrete examples of how human extinction is going to happen within five years? Why do we have specific probabilities (> 10%), but no scenarios?
I’m not even saying “AI” couldn’t cause human extinction, I just wonder why there if no better reporting on this. Because seems like kind of a big deal if it was true.
why don’t they give concrete examples of how human extinction is going to happen within five years?
Here’s one. AI is already very good at manipulating people via social media, and unfolding/combining proteins. It’s not hard to imagine an AI designing a virus that could wipe out humanity and manipulating some people to manufacture and release it. Hell, there’s already people working in AI cheerleading human extinction, so I don’t think the manipulation is even necessary. The main barrier is that it might be a while before AI can self-sustain and replicate without human help, assuming it would have an instinct for self-preservation. Once that’s solved, we’re toast.
