AJRF

Yeah, part of me wonders how much of it is shared delusion and how much is just cut throat racing for the exits.

Even if the tech + business fundamentals don't match reality, there's just so momentum in favor of a "win" here.

What we have today is interesting tech, with many, many applications - but the path to the kinds of things mentioned in this prospectus have no clear path, it seems like scaling laws are plateauing, but those invested speak as if it's a foregone conclusion we get AI we see in science fiction movies.

There is a good documentary called "All Watched Over by Machines of Loving Grace" which documents how computing started, the touted benefits (it would fix society, all illnesses, liberate humanity from work) - and the slow decline in optimism towards what we have today. I feel like AI will play out the same way.

why do you think that?

Do cloud providers have a lock-down mode where you can press 1 button and stop all spend?

Having had to use Meta Ad Manager recently, it's all starting to make sense.

If you continue down this path, your golden goose is COOKED

What were the signs for you to get checked out?

If you are interested in tech outside of consumer electronics lobsters is a good alternative to here.

not giving them a chatbot

It weirdly download an ONNX model to your browser, complete with config and tokenizer haha. First load was 50mb for me.

Am I wrong in thinking Battersea is absolute murder to travel too as well? Theres one line to a tiny station and it gets overcrowded a lot.

I started writing little "that's useful to remember" tips on my website. Bit more durable than shell history, and you can share them!

https://adamfallon.com/til.html

I do think there should be regulation. I think OpenAI specifically should be disallowed from further training runs until they can show competence.

RubyGems should sue the everliving daylights out of OpenAI for this.

i've never seen such a damning indictment of a language.

Most large tech companies are just job programmes, i've seen so much effort, time, cost go into things that are truly unimportant.

React Native gives you (with asterisks) - one "source" for things to go wrong (as it sits on top of the native implementations of the UI + Native APIs). You are writing at that source level, and that filters down to the native builds. I am WELL aware to do certain things on each platform you need to get your hands dirty, but most apps are just serialising JSON from a database and displaying it.

AI writes very buggy, sloppy code.

They've decided - instead of containing that code to 1 surface, they would now have 2 surfaces they throw slop on top of.

I look forward to the 2027 version of this where they've gone back

> leaving aside the idea that OA might've used data from the researchers Codex sessions

Why leave that aside? That is _the_ story.

If a Chinese research lab did this we'd call it espionage.

Thanks for answering in good faith.

Yeah so unverified PR's is super annoying, but people using LLMs to answer things incorrectly they could look at themselves is a big issue too.

I often hear this get dismissed along the lines of "oh you didn't context engineer hard enough" - but the default state of the model is to very confidently state a thing to be true when it hasn't searched correctly.

That is a trait of a very junior engineer - one who, if they never learned to fix this behaviour would be fired.

It seems objectively _worse_ than what we had before - trained engineers who gained wisdom over a long time horizon and had a reputation they'd lose if they kept incorrectly stating things.

I see a lot of enthusiasm from people who work alone or who have total control over a project and nothing but abject misery from people who work in a professional setting with a group of people using coding models.

I'm just building some data points.

Simon - I hope this is not a rude question, but do you work with other engineers?

Such a big disconnect in the coverage of coding models and the output they produce.

If you discard coding purity questions like style, architecture, cleanliness - the stuff they come out with is buggy & error prone.

The problems seems architectural - in that context windows are limited and you need more compute to increase them, married with the fact the models are really over confident. But if you do increase them it causes mode collapse. Yann LeCun has a really good graphic in his slides of a circle (all possible answers) and a red line coming from the centre depicting the one correct path. How do you actually stop the model going into the subsequence of wrong paths? I don't think it's possible.

I've had so many times in my day job someone has told me (Claude told them) there is a bug in my code, I look at it and nope - it just didn't look up the right file. Then you push back on it and it completely crumbles and says sorry.

I wouldn't keep an employee hired who did that over and over again and never learned

What are they currently making?

> audacious mission to make Linux on the desktop happen

and a arch linux reskin with a tiling window manager + his companies apps pre-installed is the way to do that?

I missed that, thanks for pointing out

Weird thing to see at number 3 on HN - is there some subtle context I am missing here?

Are we wink winking that it's a lot of fixes?

Who is in charge of AI strategy for the US? They are doing a terrible job!

I keep seeing people working on Passkeys get real defensive when told they don't make sense to people.

I've worked in tech 12+ years and I _hate_ when a Passkey prompt comes up, its only ever slowed me down.

But the devs who work on them are quite rabid, and keep dismissing real criticism of their implementation.

I read this as deeply embarrassing for OpenAI - they can't securely contain a program, even with their apparently amazing AI.

I don't think the parallels are quite as clear for a few reasons:

1. Lock in - with an LLM, there is practically no lock in because of the inputs & outputs being text. You can move easily

2. Motivation - I think you underestimate just how high some of these bills are for companies. Finance departments are already getting mandates to reign in spending even at the high level of subsidization.

3. Political Meddling - we're now at the point where the US strategy seems to be to artificially limit access to powerful models. If China continues its trajectory they will have models as good as Fable in 6 months to a year, and they won't lock it off. So cheaper, better models that are available is a massive incentive to switch. China is much less motivated to ratchet prices up if it's winning them marketshare. I do think David Sacks + AI strategy for US Gov are being very short sighted and it's going to blow up in their faces.

I present article A: https://x.com/asha_shar/status/2074124008795369482

> This is matching your phone's location

But what is getting that?

wait - so you think that the map is made up of people who are all sitting on that website using the Find My Train demo?

I think you are missing the point - what is collecting data on all those trains.

Few questions, wonder if anyone knows the answers:

1. So it's Trainline on a persons phone that is tracking this info and using it to enrich this service? I use Trainline and didn't know it was doing that, but I do have location permissions on because I was told that powered the search picker when I started using the app.

2. What did they use _before_ Trainline? Or was Trainline selling user location data to them?