I agree it's very frustrating to use with custom tools/harnesses that can speed up the process for domain specific purposes.

1/30,000 * 100 = .003

The fun part is that you will never know if your neural net classification project is getting silently sabotaged because their classifier doesn't work!

The most fundamental essence of what they do is exactly what you say they don't: estimate.

It turns out coding has to do with a lot of the same reasoning needed in math or in legal analysis, even if the grammatical expression is different.

This is less true of lower intelligence tasks. Classification requires a lot less reasoning capacity and so can be much smaller and more specialized.

For enterprise, Anthropic is crushing it. In the manufacturing sector I anecdotally hear a 2:1 ratio of Claude to ChatGPT for teams who are settling on a platform.

Kimi 2.6 is a 1 trillion total / 32B active parameter model that's something comparable to Sonnet. Sonnet's API pricing is $5 in, $15 out per million tokens. Deepinfra serves Kimi at $0.75 in, $3.50 out, and about the same at openrouter. So you're looking at a 4-7x multiple that Anthropic is charging compared to market rates that any plebe can get with a credit card.

This seems like a good approach to me, I work with a lot of legacy ERP-using companies in the manufacturing sector and can immediately see how we could put this to use for our customers.

I especially like that it's not doing computer use for everything which so far doesn't really seem to be working, especially outside the browser.

Mold flow simulators already exist, but they're slow. So if you're a mold engineer, and you want to try different material properties, each run takes time.

By making this into a transformer (no idea what that model looks like) it can run pretty fast, because it's not mathematically solving it, it's running a learned function approximator. So for the mold engineer, suddenly they can just change values around and get pretty fast analysis.

They appear to not be selling it as a product per se, but partnered with a German company called Simcon to sell it, whose website lists it as a preview, so it's unclear how good it is compared to conventional simulations.

The idea that exponential growth will continue with stacked sigmoids is also not a given. An example is the nail. Nails used to be about half a percent of US GDP. That's a pretty big number! A series of innovations stacked on each other (each innovation having its own sigmoid) to reduce the cost of nails. Nails dropped in cost by over 90%.

But eventually nail manufacturing reached a floor. And since the mid-20th century, we haven't gotten much better at making nails. The cost of nails actually started increasing slightly. We ran out of new innovation sigmoids, so we got stuck on the last one.

So what you actually have to predict is whether there will continue to be new sigmoids, not whether the existing sigmoid will asymptote (we already know it will).

This is much more difficult to forecast, because new sigmoids (major new innovations) tend to be unpredictable events. Not only are the particulars difficult to forecast (if they were knowable, the innovation would have already happened), but whether there will be a major innovation or not is also hard to forecast, because they are distinct and separate from any existing sigmoid trend.

So we are left with the idea that all current innovations in AI will asymptote in their scaling as they reach the plateau of the sigmoid, but there may be new sigmoids that keep the overall trend up. Or there may not be. We don't know.

That's not very satisfying, so we'll get to keep reading articles like this one.

It matches the pattern of LLMs being very good at simulating the form of work output, which is an issue with code but it seems quite exacerbated with anything non-verifiable, like written communication.

That some element of the world is dystopian means the world is therefore dystopian.

And if the world is dystopian, then every element of it is dystopian: no element of the world is not dystopian.

It's an interesting "poison the well" type of argument, but it also means that if one accepts that all of human history has had dystopian elements (chemical weapons, slavery, feudalism, etc etc), that therefore the world has always been dystopian, and that then therefore everything in the world has also always been dystopian. And you've created a label that includes everything that has ever existed and excludes none of it, which is meaningless.

I'm not really following.

A lathe operator isn’t any good if they don’t frequently operate lathes.

A articulated robot implementer needs frequent experience implementing robots to be any good.

That doesn’t mean lathes or robots are useless. Nor does it mean they have failed as products because they require expertise.

I do think it raises questions as to whether vast swathes of the population will be effective at using LLMs. Are they scissors, or a lathe?

I've noticed Claude does far fewer listicles than ChatGPT. I suspect that they don't blindly follow supervised learning feedback from chats as much as ChatGPT. I get Apple vs Google design approach from those two companies, in that Apple tends not to obsess over interaction data, instead using design principles, while Google just tests everything and has very little "taste."

In general I feel like the data approach really blinds people to the obvious problem that "a little" of something can be preferable while "a lot" of the same is not. I don't mind some bullet points here and there but when literally everything is in bullet points or pull quotes it's very annoying. I prefer Claude's paragraph style.

I suppose the downside is that using "taste" like Apple does can potentially lead a product design far, far away from what people want (macOS 26), more so than a data approach, whereas a data approach will not get it so drastically wrong but will never feel great.

Boss-man actually has a very difficult time turning legal theoretic right into actual deliverables.