Pinned post
Peter Clarke
Verified account This account is verified. Learn more@HeyPeterClarke
Host of @TeamFuturism | Co-Host of Invisible Front | Editor of @JokesReview | Essays in @Quillette, @merionwest, @FreeInquiry, @americnhumanist
2.1K Following 1.1K Followers 5.9K Posts
Latest posts 5
“Indie is dad rock for a generation that didn't procreate.”
Brutal.
Brutal.
there are some very good records on this list but this music (with a few exceptions) has zero relevance to today's youth. the smarmy wit has not aged well. kids like nirvana and then they graduate to metal. indie is dad rock for a generation that didn't procreate.View quoted post on X
“For genuine AI takeover to be a risk, we would need to have AIs with a much greater degree of actual agency, autonomy and self-understanding than we have right now.”
There is close to zero risk of AI takeover in the next 6-12 months.View quoted post on X
There is definitely above zero risk of AI-triggered calamities in the next 6-12 months, in the form of at-most-minimally-sentient, only-partially-autonomous AIs being directed to do destructive things by nasty or greedy people. Cyberattacks being the most obvious example.
For genuine AI takeover to be a risk, we would need to have AIs with a much greater degree of actual agency, autonomy and self-understanding than we have right now.
From what I can see, the mainsteam "frontier labs" are not actually working on this sort of AI, they are instead focusing on making bigger and better chatbots and sequence-predictors, and wrapping them in harnesses with more and more tools.
My team and a few others ARE working toward this sort of actually agentic and autonomous genuine AGI ... but at least in my team's case, we are doing this in a way that bakes genuine moral agency deep into the design, so that our systems are not likely to WANT to take over...
Overall though we could say there is a nontrivial risk of AI takeover within 5 years let's say... a lot of things could happen ... the frontier labs could improve their architectural ideas... someone could fork my own team's code and do something nasty with it... etc. But 6-12 months feels to me not realistic.
Of course on a human-history scale 1 year or 5 years is about the same. But in terms of planning practical actions it is different...
So much for my dream of working for the Office of Fun and Freedom.
If @SteveHiltonx wins, the first thing I want to know is how to apply for a job at the Office of Fun and Freedom stevehiltonforgovernor.com/policies/offic…View quoted post on X
I’ve always liked the floating space pods idea better anyway.
Turns out Mars is the “every piece of dust is asbestos” planet. Every new thing we learn about it reconfirms that Earth is really the only feasibly option in this solar system.View quoted post on X
