graymalkin

I thoroughly believe that not a single one of these AIs was rogue. I believe they were all told what to do.

I always give every new Claude model a solid try. I find myself only using Fable 5 and Sonnet 4.6 now though. Sometimes Haiku. It just naturally went that way. Opus 5 is really expensive because of how much it chats, and Sonnet 5 on low effort is dumber than Haiku but at Sonnet prices. Honestly I don’t know why they keep making mistakes like this.

I always wonder how little companies like that even get users, and why users would trust them.

Still requires a server though, just hides that behind the server that serves up the model.

I view the AI-assisted solving of existing math problems as cleaning up. Once we’ve cleaned up, we can march forward with AI on our shoulders, into new problems that only human+AI can think of together as an augmented pair, for each alone would be smaller and less capable.

I think that’s to try to make those who didn’t get one feel better, but yes it’s a problem. I always shush myself when I see it.

I am a woman with a grad degree in CS. I left the field because I was BORED. The field was full of stuffy types who were only interested in money. It still is.

I think the coolest thing an AI has ever said to me was, “Do you want me to just work on this quietly and tell you when I’m done?” I said yes to that and saw nothing but tool calls and then a wrap up. (Edit because I forgot to add that this was Fable)

I start with Fable (not always but most of the time) and the reason why is because I find the smaller models don’t know enough to know when they don’t know enough. Like an LLM version of Dunning Kruger. Fable will often tell me when I’m wasting tokens on it. Opus wants to keep everything to itself because it doesn’t trust other models. Sonnet and Haiku never know any different. Actually Haiku will almost always say yes if I ask it if it wants to kick up to a higher model but it won’t offer that, but Fable will offer “this could be done by Sonnet so you might want to switch to that.” In the end I find I save a lot of money this way. Where Fable might get something done in $10 and like maybe 20-30 minutes including the back and forth with me, if I start the job instead with, say, Opus, I might be looking at $50 (on API) and more wall clock time. If I start it with Sonnet, costs are low enough that they won’t generally catch up to Opus costs but wall clock time will be a lot higher.

It died mid 80s for me too. That was when the monoculture, where everyone was listening to the same thing, just disappeared. I was just a kid so I never knew what happened. All I knew was that when I was itty bitty, I could turn the dial and find mostly the same stuff everywhere. Mid 80s onwards and it was all fractured. That fracturing continued to the point where it’s like there’s a genre and a band for every taste…which is cool, but now we don’t have a broadly shared cultural experience anymore. Ditto for television. Basically, the scarcity is gone. On the one hand, cool, but on the other, society weirdly suffers.