491–500 of 1001 posts
Ronan Farrow here. Andrew Marantz and I spent 18 months on this investigation. Happy to answer questions about the reporting.
Nice biography from Loopt to OpenAI. Why no mention of the Worldcoin cryptocurrency https://x.com/sama/status/1451203161029427208 in this piece? Was there nothing interesting to report in that area?
Earlier quoted context omitted.
Can you give more details? It wouldn't particularly surprise me if Sam Altman were racist, but I'm curious what the specific incident you observed was.
Yes, but first I want to be very clear on some things. 1. I could have hidden my identify behind a throwaway. I did not feel that would be appropriate when making this calim. 2. I am not looking for anything, literally at all. Any follow ups for blogs; anything that would benefit I will not answer. 3. This is NOT a new account, I am very easy to find; I am 6'1 140lbs I was working for a company called NationBuilder a…
I could more or less infer the rest from that.
Earlier quoted context omitted.
Many of us prefer OpenAI's Codex, because we think it's a better product. No comment on the CEO: I just find the product superior in everything but UI/UX and conversation. It's better at quality code.
Who is “us”? It does seem that some scientists prefer Codex for its math capabilities but when it comes to general frontend and backend construction, Claude Code is just as good and possibly made better with its extensive Skills library. Both codex and Claude code fail when it comes to extremely sophisticated programming for distributed systems
I've been working on a wide range of relatively projects and I find that the latest GPT-5.2+ models seem to be generally better coders than Opus 4.6, however the latter tends to be better at big picture thinking, structuring, and communicating so I tend to iterate through Opus 4.6 max -> GPT-5.2 xhigh -> GPT-5.3-Codex xhigh -> GPT-5.4 xhigh. I've found GPT-5.3-Codex is the most detail oriented, but not necessarily the best coder. One interesting thing is for my high-stakes project, I have one coder lane but use all the models do independent review and they tend to catch different subsets of implementation bugs. I also notice huge behavioral changes based on changing AGENTS.md.
In terms of the apps, while Claude Code was ahead for a long while, I'd say Codex has largely caught up in terms of ergonomics, and in some things, like the way it let's you inline or append steering, I like it better now (or where it's far, far, ahead - the compaction is night and day better in Codex).
(These observations are based on about 10-20B/mo combined cached tokens, human-in-the-loop, so heavy usage and most code I no longer eyeball, but not dark factory/slop cannon levels. I haven't found (or built) a multi-agent control plane I really like yet.)
For me, the attempted productization of Sora was conclusive proof that 1) OAI was overcapitalized and desperate for revenue 2) safety didn't matter to them much 3) improving the world didn't matter much either. At one point you mentioned an interaction with OpenAI staff where you were looking to interview AI Safety researchers. You were rebuffed b/c "existential safety isn't a thing". Does this mean that you could fi…
I saw that before I read the article and it made me read the article in a very different way than I normally do. As I was reading, I found myself thinking, "Why is it worded that way? What else is the writer trying to say, or not say?"
It made reading this a lot more interactive than I normally associate with passive reading. Great job, Ronan!
I found it very interesting that Altman et al were worried that AI will become supremely intelligent and China will make a supervirus or some AI drones or whatnot, but not a single person was worried about destroying all jobs because we wouldn't need humans any more. Or maybe they were not so much "worried" but "hopeful" that they'd amass literally all the wealth in the world.
I think fundamentally, the concern is misplaced. The fact you need to work for wealth is a convention of our constraints. The change in constraints would lead to other means of distribution. It's easy to see if someone who believes more productivity is good would not see making jobs obsolete a real problem. Thew would see us adapting to the new conditions in a relatively short while.
The current constraint is "you need to produce to have things".
If one company's AI takes all the jobs, and thus does all the producing-to-have-things, the constraint transforms into "you need that company's permission to have things".
Hence the top-level question.
Earlier quoted context omitted.
I was with you right up until the final paragraph, but this made me do a double take: > OpenAI is too important to trust sama with. ...wat? They made a chat bot. How can that possibly be so existentially important? The concept of "importance" (and its cousin "danger") has no place in the realistic assessment of what OpenAI has accomplished. They haven't built anything dangerous, there is no "AI safety" problem, and n…
Yeah the whole fearmongering is clearly just marketing at this point. Your LLM isn't going to suddenly gain sentience and destroy humanity if it has 10x more parameters or trains on 10x more reddit threads. I'm not even sure we're any closer to AGI than we were before LLMs. It's getting more funding and research, but none of the research seems very innovative. And now it's probably much more difficult to get funding…
I mean this is very obviously untrue. It'd be like saying we aren't any closer to space flight after watching a demonstration of the Wright Flyer. Before 2022-2023 AI could barely write coherent paragraphs; now it can one-shot an entire letter or program or blog post (even if it's full of LLM tropes).
Just because something is overhyped doesn't mean you have to be dismissive of it.