samvines

![Soon, at each new model of AI along the current capability curve, you will start to see large discrete jumps in ability in economically important areas, because the previous AI ability level in some aspect of the job just wasn’t good enough and bottlenecked progress. When bottlenecks are released, it looks like a leap forward.

It is going to look like unexpected gains in AI capacity, and, indeed there is no sign that the current exponential ability curve is slowing down (so far), but it is going to be like what happened in coding: as soon as models crossed a certain threshold (with Opus 4.5, GPT-5.2, and Gemini 3), suddenly Claude Code & Codex were viable. Before that, it was all about coding assistance, afterwards it was all about agents from despite relatively small gains in model ability.](awful.systems/…/3eb7c440-aa86-4a4b-b0b8-c548fb45b…)

There is just something so inherently smug and annoying about Mollick. He is one of those low information boosters whose posts sound intellectual until you really think about them.

Tell me more about how the pile of cursed spaghetti that is Claude code is now viable due to model breakthroughs. All I see are hype men saying “the new model is a team of PhDs in your pocket” and then releasing disappointing updates or saying “the new model is too dangerous” because they have some vaporware powered by human crowdsourcing.

Also coding is not like other areas - you can test for hallucinations by compiling and printing and running tests.

I guess my first mistake this morning was opening linkedin

AI bros are seizing the means of computation: RAM, GPUs, SSDs and now HDDs… I don’t think there’s an actual conspiracy, just lots of MBAs following their noses towards the $$$. That said, time to buy a new lipo battery for that 10 year old laptop in the loft and stick Linux on it - before the lithium miners announce they’ve sold the next 12 months global supply of Lithium to Altman because he needs it to sleep at night…