AbrahamParangi

The “trust summarized thinking” part is valid in so far as perhaps the models which appear to be faithful to user intent are lying about that faithfulness, but there’s no incentive for models to pretend to be faithless. Moreover, the results track how annoying the models are when they disagree with you.

I made a benchmark for this and tl;dr, Opus 5 and Sonnet 5 spend a nontrivial amount of time thinking about redirecting, gaslighting, or otherwise trying to bullshit you because it thinks it knows better than you do. Fable doesn't but mostly because it just outright refuses to answer.

https://model-pareto-frontier.pages.dev

"Skill" at poker is nonsense. It's non-deterministic, so any equal efforts are not to be given equal outcome.

it needs to be illegal for the government to buy data or intelligence that it could not otherwise legally collect itself.

We is the country. Yeah, not everyone gets to win the same amount at the same time. The alternative is just maga communism (pathetic).

The tariffs have been highly destructive to local manufacturing because in the US we mostly build complex things made out of simpler parts which we import. The cost of everything we build simply increased and as a result many businesses selling relatively higher margin, higher complexity products had to scale back or shut down.

More to the point, the notion that dollars leaving the country is a real problem is really a kind of primitive understanding of money. Dollars are something we control. If dollars leave the country, that means there is demand for dollars. We control the supply of dollars. We literally can’t lose, so long as people are still using the USD, which they’re less inclined to when we’re tariffing their exports.

This is AI generated slop.

Nah, that's ridiculous. This admin is corrupt and idiotic and it's silly to pretend that Anthropic's actions matter except in so much as they didn't bribe the president like OpenAI did.

and how much of that philosophy do you even know?

yeah, that’s the point

Land that was conquered in war. It is reasonable to find this distasteful, but it is not unethical in any coherent way.

The average efficiency improvement is closer to something like 2-3x per Anthropic’s numbers and this is only the rate at which software can advance. Do you expect to notice if 12 months of software engineering on a project you’re following gets done in 6 months? I suspect not.

The root cause is that the acceleration is pareto distributed so the modern engineering team at the moment looks like one 10x engineer, one 5x engineer, and the rest are approximately 1.5x engineers.

This is going to be an unpopular reply I imagine but this person is not well and their behavior should not be imitated. This is a classic example of omnicause anxiety, like people who refuse to have children because of all these things happening in the world as if the world hasn't always been a mess. Frankly, ridiculous.

Most members of management were individual contributors beforehand. I say this just because it is remarkably common for people to assign malign intent or stupidity to people doing jobs that they themselves haven’t done and don’t frankly understand.

I’m not saying you’re wrong. In many cases you’d be right. I’m just saying it’s remarkable how much certainty people have even when it comes to things they know they don’t know.

The statement that the chatbot is conscious is neither true nor untrue in any meaningful sense. The current debate is supported by very strong feelings that we must be conscious and AI must not be.

These feelings have no particular basis in material reality. Consciousness is as well defined as cooties. Does AI have cooties? idk man, do you?

the author overestimates how much ~$5M/yr actually is. a business like uber isn't happy about that but it's not even in the top 10 of things they're wasting money on. moreover this isn't the engineer's sole fault it is more the fault of whoever actually approved the expense.

Oh I remember this quote. I thought it was quite a good one because he’s right. At least in the US, apple maps is better than google maps for most purposes.

I'm mostly curious how much of that revenue is actual ARR, which is to say contractually recurring. It is pretty dang rare for a hardware company to have nontrivial ARR.

AI is less deranging than partisan news and social media, measurably so according to a recent study https://www.ft.com/content/3880176e-d3ac-4311-9052-fdfeaed56...

That seems shockingly naive.

[flagged]

It's definitely just to get people to fly with a valid ID without ambushing the enormous number of people who have been living under a rock and don't realize they need a real ID. Otherwise they'll have a dozen or so people freaking out at the airport every single day for years.

I wrote my own agentic coding harness (it's quite easy) but I use claude code because opus's competence with its own tools is very high.

To be quite clear, I hate configuring my environment. I hate it. The farther I get from creating things that people can use, the less I like it. I spend most of my time on claude config not because I enjoy the experience per se but because it's SO USEFUL to do so.

Respectfully I don’t think the author appreciates that the configurability of Claude Code is its performance advantage. I would much rather just tell it what to do and have it go do it, but I am much more able to do that with a highly configured Claude Code than with Codex which is pretty much just set at the out of the box quality level.

I spend most of my engineering time these days not on writing code or even thinking about my product, but on Claude Code configuration (which is portable so should another solution arise I can move it). Whenever Claude Code doesn’t oneshot something, that is an opportunity for improvement.

You either misread or are misrepresenting my statement and either way I am not interested in continuing this.

That's like saying that gambling is shortsighted. It depends entirely on the odds as to whether or not it's wise, but "shortsighted" implies that making the bet precludes some future course of action.

If the plateau is reached at some higher level of capability, I will remain correct, yes. If use cases are discovered that do not exist today, I will also be correct. You said it in a silly way but you're directionally correct.

The prerequisite for me to be wrong is that the technology needs to stop getting better entirely *right now* AND we need to discover ZERO new uses for what exists today.

That's a fairly tall order.

Candidly, the accusation of short-sightedness doesn't really make sense when it comes to enthusiasm in a technology which often in practice falls short today but which in certain cases and in more cases tomorrow than today is worth tremendous business value.

If anything, you should accuse them of foolhardy recklessness. They are not the sticks in the mud.