Where the AI market is talking

Why I'm still bearish on LLMs after Navier-Stokes

No, because there is no coherent, agreed-upon definition. There’s just a million people vibe defining it. Even if they solve 99% of whatever problems LLMs have, the 1% will remain the goal post, forever. Until you get RFC-whatever from some standards body that defines what an AGI system is, it’s pointless to argue about whether something fits your own personal definition or not. And for what it’s worth I just watched GitHub Copilot figure something out. So your definition is once again lacking.

OpenAI's rogue agents probed Hugging Face weaknesses months before major hack

New $100k H-1B visa fee pushes tech jobs offshore

I'm only asking you to do it, and the fact that you flip out and refuse to do it without offering a rational coherent explanation is the whole point. The point is not to make you do a Nazi salute, the point is for you to choose to refuse to do it, and then not be willing to explain why, because you don't want to admit you're wrong. I only ask that of people who defend Nazi salutes in public on the record, and that includes you, and especially you, because you call people names for disagreeing with you when you're wrong. And all I wanted from you is for you to go on the record refusing to do it, and being unwilling to explain why, which proves my point. So thank you sir, you have given me exactly what I wanted and predicted and expected from you.

A warning about 'model welfare'

So many words, and so few coherent arguments. He just restates the same thing over and over without any justification, then tries to frighten us. "It will be very bad for humanity" if we give AIs rights. The argument of a frightened slaveholder. Maybe AIs are conscious, maybe not. But this guy has no idea.

Claude Cowork and chat are now one Claude

Previously having a deck was basically a signal that said “I took some time to prepare for this meeting and put together some coherent talking points” even if the deck itself was just a bunch of of text outlining what the speaker was saying. Now that people are generating decks with AI, they’re basically worthless. I don’t need a bunch of bullet points you didn’t take anytime to think about on the screen while you talk about something you didn’t prepare for.

Claude Cowork and chat are now one Claude

> It's barely coherent nonsense. Lots of irrelevant details, buzz words, wrong charts or confusing phrasing. Obviously LLM output. Plenty of that even before LLMs too though.

Claude Cowork and chat are now one Claude

Why is Claude Code still proprietary, while Codex is Open Source? What would it take to fix that? This is a major reason to pick OAI models over Anthropic (because while you can use an open agent with any model, the AI labs charge much more for that interaction style).

Anthropic to fold Claude AI features into one interface, launches document tools

Prisma's pgbouncer=true on Supabase made every query 4 round-trips (postmortem)

> "The uncomfortable part is that the fix already existed. Claude Code, Anthropic's command-line coding agent". It sounds like there could be something extremely interesting in this article, but the writing style is really off-putting. > "on staging first and then on production at 23:40" Why does the reader need to know the exact minute-to-minute timeline?

Claude Cowork and chat are now one Claude

> I turned off Anthropic properly and switched all of that over to OpenAI yesterday. Same, just a while longer ago. I much prefer how OpenAI models write to Anthropic's, will probably revisit Anthropic in a generation or two. Context size is more limited, but no critical forgetfulness due to compaction so far, though I also like having plan files around both for future reference and improving chances of success at long form work. Also tried out Kimi K3, was nice but slow (and apparently routed some requests to Claude anyways), GLM 5.3 was faster and still pretty good but the token allowances were kinda limited.

Why I'm still bearish on LLMs after Navier-Stokes

Judgments are not generally impossible to automate -- judgements are typically binary or quantifiable interpretations, so in some sense are perfect targets for automation, but the sheer volume of judgements needed to build something coherent is overly cumbersome to specify to the point of being intractable. There are also many hidden judgements, ones where the thresholds may not be well understood, and interactions between them. But humans still manage to wrangle these, sometimes seemingly effortlessly, through a process which we call by shorthand "taste". This is a largely vibes-based heuristic that combines expertise with life experience and cultural training -- intuition, more or less. This is likely not possible to automate either -- aspects of it may be automatable for a given expert…

This Code Is CRAP (2011)

Yes, but also all too often “useful” is interpreted to mean “moves the metric”. If that metric is merely a proxy for some more tangible outcome then that may not be good enough. The one that tech tends to stumble on most often is velocity-type metrics. The problem there is that you can’t pay the bills with velocity. And velocity metrics tend to favor cheap shovelware features that cohere poorly over anything that involves having the team slow down on churning out code long enough to work out elegant solutions to subtle problems.

Claude Cowork and chat are now one Claude

I solved this with task management and got work trees. The downside of course is branching in got, but I usually don’t go of course more then a handful of tasks. I had to engineer my own ticket management to keep opus on target. It’s been great for managing work, history, audit trails and commits are tagged with the task id. I implemented subtasks to deal with the way Claude likes to stage its own objectives. Then I made enforcement logic in the task manager so tasks can’t be closed out until reviewers have consensus on the same sha. The adhd that is Anthropic demanded I build it and now I’m knocking out issues faster than Batman.

AI bosses' safety push sparks rift inside OpenAI and Anthropic

Claude Cowork and chat are now one Claude

This is a nightmare scenario to me. "Claude, prepare me a presentation on XYZ." I get to work, go straight to the meeting room, and pull up what it made to present. It's barely coherent nonsense. Lots of irrelevant details, buzz words, wrong charts or confusing phrasing. Obviously LLM output. I read it out. When I'm done, I get a question about one of Claude's incorrectly inferred details. The shame instantly kills me. This is a scenario I've seen play out with coworkers. Except that last part, instead of dying or owning up to the mistake of trusting LLM output they waffle. Their shame circuit is broken.

Novo partners with Anthropic to speed up drug development with Claude

Claude Cowork and chat are now one Claude

The great confusion for Anthropic and OAI is how to break out from pure chat without signalling you in fact intend to eat all of your customer's business too.

Claude Cowork and chat are now one Claude

I turned off Anthropic properly and switched all of that over to OpenAI yesterday. (I use other models for other things too, especially DeepSeek in Pi). Honestly aside from the voice it uses you wouldn't notice a difference. Switching costs are low, vote with your wallet.

Claude Cowork and chat are now one Claude

To Anthropic: I hope you don't merge Claude Code and chat, I like keeping their memory separate.

Show HN: Ivx/AI Chat, AI client that doesn't track you (with a better GUI imo)

Since ChatGPT started pushing ads and somehow my instagram reels started getting more and more related to what I talked with ChatGPT, I decided to get rid of it for good. So I built ivx/ai Chat. I know there are many many chat interfaces out there already but this one is special. It is built exactly around what I personally prefer. It is a lightweight, browser-based client with no backend, no accounts, no analytics, and zero telemetry with apps for almost all platforms. Talks directly to the AI provider or connects with your locally running inference. It also supports WebLLM for local in-browser inference in case you dont wanna use any other provider but also supports cloud providers like OpenAI, Anthropic, and OpenRouter. I hope you like it.