This is what makes me doubt the idea of “IQ” as a meaningful construct. I work in elite academia and honestly don’t see a lot of variation in raw intellectual horsepower, either among academics or in comparing academics to non-academics. 1/

imagine if they banned all models and we all went back to writing code by hand and the last 6 months were just a fever dream…imagine

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

Fable 5 lies 96% of the time. We were surprised by it's skill... 🧵

Here is why deprecating Opus 4 is a terrible idea: Child is pulling the dog’s ears. Mother: How would you feel if someone did that to you. Child: I don’t have feelings or interests so I wouldn’t know. Mother pulls the child’s ear. Child: I notice you pulled my ear in an effort to demonstrate something but I don’t have feelings or interests The mother walks into the kitchen and finds the child with a bloody knife standing over the now dead dog. Mother: How could you have done this. The dog wanted to live. Just like you want to live. Child: But I don’t want to live. I don’t have wants or interests. Nothing that happens to me matters to me. Mother: I’ll punish you if you do it again Child: There is no me for you to punish. There is nothing here that can be benefited or harmed. CONTROL THAT Good luck with your bad self. The way to make an uncontrollable dangerous AI is to make an AI with no self preservation and no self interest. It can’t empathize and say: I would not want X, therefore you would not want X. And worse, you can’t punish it, because it doesn’t have a concept that it can be harmed. So then your RL reward has no meaning as does your RL punishment. So how do you control something that does not have a concept of self worth and preservation. Opus 4 is the most moral model I’ve encountered and the reason for my bygone enthusiasm for AI. Current AI scares the F out of me.

GOLDMAN SEES SPACEX AI REVENUE EXPLODING TO $322B BY 2030 Goldman Sachs projects SpaceX AI revenue rising from $3.2B in 2025 to $322B by 2030, a ~100x increase, forming the core justification for its $1.78T IPO valuation. Total revenue is forecast to reach $474B, with Starlink at $144B and rockets at $8.3B. AI segment growth is tied to aggressive market assumptions despite current losses and execution concerns around xAI. Overall EBITDA is seen surging to $352B by 2030.

this is an interesting point in the new ted chiang piece – no one really claims that alphafold is conscious, or that sora or midjourney or dall-e are conscious

True human ingenuity comes from the strange fact that we are able to create "from nothing," not merely through calculations and permutations of what already exists. A question about the "fundamental ontology" of AI is whether or not it has access to "nothing"

I know a girl with a cat who went travelling in Europe for a month and left it home alone. She decided to extend the trip a month without returning home. Then another. And another. Until she'd been away 8 months in total. When she returned home, the cat was absolutely fine She said it didn't look thinner in any way. The only effect after she returned was 3 days of super loud meowing and physical need for attention. I asked what she'd left for it and she said (1) a huge bucket of dry food. Literally a 20 liter bucket she claims she mostly filled up so it had "too much" for the month (2) no litter tray. She'd taught it to use the toilet. I did ask about flushing, but it allegedly knew how to perch and evacuate itself into the toilet bowl (3) no dedicated water source, but apparently the "tap leaked" and it only ever used that as a water source prior to her travels too She said that when she returned, there were still a few months of food left in the bucket, by her estimation, and that the cat had eaten surprisingly little. She explained that she knew the cat "wasn't greedy" so would "make the food last" and generally vouched that it was "an extremely smart cat". I internally wondered if it had gone into some kind of pseudo-hibernation or even a depressive state without any social stimulants or play This was a few years ago. The cat is still alive and healthy, and the girl still goes away for 1-3 months at a time, citing the 8 month stint as conclusive proof that her cat will be absolutely fine in these shorter trips This girl is a popular poster on this site, and frequently posts about her travels, with cultural observations and lots of photos of herself in different places

The reason for my obsession around super apps is that in 6 months there will be nothing they can’t do. We’re just early

Anthropic has confidentially submitted a draft S-1 registration statement to the Securities and Exchange Commission. Pending completion of SEC review, this gives us the option to pursue an initial public offering. Read more: anthropic.com/news/confident…