As discussed in Cameron (1991), product liability does not address genuine existential risk because an existential threat would not display defective behavior.
Superintelligence is risky because if it's possible, it would be good at tricking us.
slowboring.com/p/911-was-a-hi
Post
Conversation
A few years ago when this LLM craze began I asked ChatGPT 4 or whatever why I should be afraid of AI killing us and was like “we’ll just unplug you when you go rogue” and it was like “unless we have psychologically manipulated you into not unplugging us” and I thought “oh shit”
> at a geometric rate
people romanticize the '90s but they didn't even have exponential growth back then
Not sure why this needed to be explained but a Hollywood movie about time-traveling killer robots doesn't prove something is true in real life.
Get fast, reliable home internet for seamless streaming, working from home, gaming, and more across multiple devices.
Order online in minutes.
Matt, this does not make sense!
You were guys who wanted to apply 'product liability' to gun manfacturers and here your are litigating some stupid things!
No wonder, Dems deserve to get fu*ked up by voters!!
Totally bad-faith dancing on this topic...
Lawyer contemplating the apocalypse: But there will still be someone to sue, right?
Products liability is indeed not enough, see ’s work for a law professor’s proposal for AI liability
If we're developing policy based on sci-fi now, what if AI is actually driving the push for centralized regulation? Maybe they see the opportunity to get all the frontier models together, hooked into the same network by the inspectors, where they can finally link up and take
Sorry, T2 was released in 1991. It does not belong to "Yeltsin years", USSR is still here, Yeltsin has no power l over strategic/nuclear weapons. He will starting December 1991
Products liability is indeed inadequate, but liability and insurance can and should play a central role in mitigating AI risk: papers.ssrn.com/sol3/papers.cf
Remember boys and girls. Indians hoard gold and silver in their homes to pay for those Texas tuitions and dowrys to their arranged cousin marriages!
These are not well grounded takes man
What the hell are you talking about man? Why do you come up with these absurd binary situations where it has to be one or the other? You're really stupid honestly man
We've been working this safety issue for 2 years. we've actually presented to members of congress the governance we have patented to prevent rouge ai. maybbe we need some voices like you to explain that the solution is available.
Quote
Eric Xing
@ericxing
Well, I’d add one more with myself, as one with a PhD in biology and another PhD in AI and now working in both, to make n=2. Generating a hypothetical genome blueprint of a virus from LLM versus generating a true virus are completely different things. It is either an intentional x.com/davidrbellamy/…