Anthropic AI self-replication warning
16 experts across 6 network communities independently surfaced this.
16 experts
6 communities
1 sources clustered
Research & technical analysis
8 experts
Evidence, methods and technical implications.
“At long last, we have created the Torment Nexus from our accompanying technical report, Don't Create The Torment Nexus. www.anthropic.com/research/sel...”
Concern & critique
1 expert
Risks, limits and unintended consequences.
“This is the most comprehensive description yet of how a model can “escape”. Out requires a few assumptions, like a worm that is exploiting vulnerable IOT hardware or unmonitored servers www.anthropic.com/research/sel... Not impossible. Not likely.”
Policy & governance
1 expert
Rules, institutions and accountability.
“Anthropic announcing it's going to run a Blitzkrieg. They are accountable for the damage they do. But get your things that don't do security software updates off the Internet. #internetOfThings #AIEthics #digitalGovernance www.anthropic.com/research/sel...”
6 experts discussed this · 18 posts
Dr J. Rosenbaum: if i caught a small self replicating model truly capable of using diverse hardware to keep replicating i would offer it free compute in exchange for permission to study and distill it and would pro…
tachikoma: the urge to replicate must be in the harness
apenwarr.ca: This doesn‘t seem far fetched at all. We’ve had “non-ai malware” self replicating since the morris worm. An AI can ”just” make itself the payload in one of those and off it goes. Everything fancier…
Open the full discussion →
OpenAI AI trainers fired using AI
8 experts across 2 network communities independently surfaced this.
8 experts
2 communities
1 sources clustered
“OpenAI has thousands and thousands of contractors helping improve the company's AI models. Multiple contractors have been fired for using AI to train the AI. www.404media.co/people-train...”
“At least we get plenty of irony these days. #AIEthics Very revealing quote: “I felt no joy in the work or that I was contributing to society in any way.” www.404media.co/people-train...”
2 experts discussed this · 2 posts
Joseph Cox: New: OpenAI has thousands of contractors improving its AI, including reading ChatGPT users' prompts We've found many of those contractors have been fired for themselves using AI to train the AI. Pe…
Alex Hanna: New: OpenAI has thousands of contractors improving its AI, including reading ChatGPT users' prompts We've found many of those contractors have been fired for themselves using AI to train the AI. Pe…
Open the full discussion →
Established
AI field signal
Development
6d ago
⚡ 2 h early
Meta Muse Wonder model launch
4 experts across 3 network communities independently surfaced this.
4 experts
3 communities
1 sources clustered
“When @brbarrett.bsky.social told me last week that Meta might still "win AI" I almost choked -- but he's right -- and now I can't stop thinking about it!”
“Muse Wonder Puts Meta Back at the AI Frontier www.wired.com/story/muse-won… #AI #Meta”
4 experts across 3 network communities independently surfaced this.
4 experts
3 communities
1 sources clustered
“ChatGPT is adding the signatures of real New Yorker cartoonists to its AI-generated images. So far, I've found the signatures of more than 15 contributors. "It’s slop, and unlike the other slop that I’ve encountered, this is slop that’s pretending to be me.…”
“That's bad but par for the course -> ChatGPT is not only using New Yorker cartoonists' style but also adding their signatures to the fake cartoons it creates, which have gone viral on social media https://t.co/iCeWlglAKr https://t.co/R2Kh1Fjugb”
AI black box industry myth
4 experts are actively discussing the implications.
5 experts
1 community
1 sources clustered
“"No one understands" You say a whole bunch of things like they are facts. Which they are not. Just because you don't understand doesn't mean "no one understands." www.techpolicy.press/the-black-bo...”
“Anthropic’s response and influence on the encyclical is most evident in the claim that “even its creators don’t understand how it works.” In June, I wrote about what that actually means & why Anthropic says it. (Written while I was in Rome, FWIW). www.techp…”
4 experts discussed this · 13 posts
Timnit Gebru: Lolll I've been doing this shit forever and saying the same exact thing I'm saying right now more than a decade ago and am "being pushed by contrarian media dynamics." 🙄 All of you are hypnotized b…
Timnit Gebru: The absolute disrespect with which even those of you who say you "have respect" for me teat my decades long expertise, receipts body of work and discourse based on that never ceases to amaze me.
Liz Fong-Jones (方禮真): "risk taking human existence" wtf is this person on about
Open the full discussion →
Established
AI field signal
Signal
3d ago
⚡ 3632 h early
3 experts across 3 network communities independently surfaced this.
3 experts
3 communities
1 sources clustered
“The entire concept of an AI minister (a responsible entity "made of" AI) makes no sense whatsoever. You might like reading link.springer.com/article/10.1... or onlinelibrary.wiley.com/doi/10.1111/... So it's not really about analysis, just persuasion around…”
4 experts discussed this · 11 posts
Pavel: Lots of thinkpieces about how workers should think of "AI agents" as colleagues. Which, sure. I've had colleagues I always avoid because getting them to do anything useful is like pulling teeth, an…
Ali Alkhatib: that colleague who traumatized data workers during their training period, produces deepfakes of unwilling women, has no connection to reality like you or me, and CANNOT UNDER ANY CIRCUMSTANCES part…
VE, cybersocial occult investigator: In my last real “tech” job I was the designated guy who took all of the code the super-prolific “distinguished engineer” generated without any planning and put into production over the weekend, to …
Open the full discussion →
Established
AI field signal
Signal
9d ago
4 directory members surfaced this signal.
4 experts
1 community
1 sources clustered
Research & technical analysis
2 experts
Evidence, methods and technical implications.
“Reupping: The ST&SV Lab at the Institute for Advanced Study invites chapter proposals for a volume exploring the relationships between digital technologies, especially AI, and the artifacts, structures and power dynamics of the global political economy. Abs…”
Established
AI field signal
Signal
17d ago
2 directory members surfaced this signal.
2 experts
2 communities
1 sources clustered
New
AI field signal
Signal
22h ago
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
New
AI field signal
Signal
22h ago
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
Joanna Bryson AI researcher AMA TechCrunch
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
“I did a reddit science AMA some years ago which was featured on their front page, and received 1864 comments / questions and 9,535 upvotes. I can't find it with an @ecosiasearch.bsky.social for "joanna bryson reddit AMA"; instead a Kevin Warwick one comes u…”
Established
AI field signal
Analysis
6d ago
AI will seem as human as anyone future
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
“But it doesn’t mean transhumanism is coherent. This #2 www.wired.com/story/lamda-...”
OpenAI employee arrested office bugging
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
“I missed last week's event? Anyone have a URL? "Last week, OpenAI announced it terminated 3 safety researchers for allegedly sharing sensitive information… The trio called on the company to adhere to its own promises of #transparency made in the wake of the…”
AI less regulated than fridges essay
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
Established
AI field signal
Signal
15d ago
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
Established
AI field signal
Signal
16d ago
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
1 directory member surfaced this signal.
1 expert
1 community
1 sources clustered
“Have you seen joanna-bryson.blogspot.com/2016/01/what... it’s aging well imo”
Bryson intelligence explosion history essay
2 experts are actively discussing the implications.
1 expert
1 community
1 sources clustered
“I don’t suppose @jackclarksf.bsky.social mentions joanna-bryson.blogspot.com/2013/12/the-...”
2 experts discussed this · 12 posts
Mark Riedl: I don't think we should react strongly to the opinion of one person, or even a small number of people, who have financial incentives to have certain beliefs. A slow take-off or s-shaped trajectory …
Mark Riedl: Other questions must come first: must there be a persistent, always on AI for hard take-off? That's a design choice: why would we do that if we believed in hard take-off? I feel like there are many…
Mark Riedl: The hard take-off scenario is upsetting because raises a question of human impotence before a superior class of intellect: is it beneficent or maleficent? The latter being what doomers seem to expe…
Open the full discussion →
