> we can fairly say that Hinton was wrong because his prediction did not reflect what a radiologist actually does: had Hinton bothered to ask the question, he would have learned that practitioners do much more than interpret diagnostic images.
This has been the crux, for me, of arguments I have had with non-technical people about why, as a software engineer, I am intensely relaxed about AI "taking human jobs". Recently Ford quietly started re-recruiting engineers after they discovered the hard way that their engineers had been doing more than they thought they had, and I sent articles about it to friends I had previously been discussing the matter with. It is still mind-blowing to me that a business historically famous for automating manual labour can have made this mistake, i.e. of literally not knowing what their staff actually do.
Humans aren't machines, who would have guessed? But, perhaps more importantly, machines aren't humans, and the adherence of a lot of domain experts (that should know better) to fairy tale fantasies about the tech they are literally building genuinely makes me wonder if the whole field has just gotten dumber in the 21st century.
It's said on every one of these but it bears repeating: existing cybercrime legislation already covers this - "rogue agent AI associated with OpenAI attempted to hack xyz" = OpenAI attempted to hack xyz.
A summary of a research article isn't a research article. Do you think newspapers are scholarly journals? Also, you do understand that anyone can just upload a document to SSRN, right? "Preprint" doesn't actually mean anything academically. Also, the methodology of that paper is absurd: Google trained a model and then asked that model what it thought about AI policy. Can you not think of any way bias might be less than controlled for doing it this way?
That's it yeah. These kinds of pieces have been written for centuries. Surprise: people like what they are familiar with! There is no meaningful conclusion to draw from this. Look at Monet or Satie. In their own time? Grotesque, offensive - the kind of art scruffy pompous intellectuals enjoy. Now, a century later? At home in a greige IKEA showroom house being sold on the open market. To be clear: I don't think there is anything wrong with this - I love IKEA! I am buying a greige house as we speak! It is just not an interesting observation, and when writers attempt to use it as an argument against making new things, it's facetious at best, and at worst? Well, several interesting possibilities.
> Why are we commissioning styles that we say we dislike? There are several interesting possibilities.
Like what? A shadowy cabal of liberal elites pulling the strings from the shadows? Stripe (who own Works In Progress) founder Patrick Collision donated to the Trump campaign in 2024. Tiresomely obvious what this "piece" is trying to do. Find a new gimmick imho.
I'm vibe coding a toy chatbot with OpenCode and GLM 5.3 - I'll open source it once it's meaningfully done (keeping the repo private for now until I am _so so_ sure there are no secrets in it!). This has been an exercise in learning about lots of new AI SWE concepts including some tech (Mastra, LangSmith, DeepInfra) - as well as a chance to try some open weights models. The actual useful output is an iteration loop orchestrator based on GitHub issues/PRs which I have found a lot of fun to use. That'll also be open-sourced.
Benchmaxxing is the word you're reaching for here (though code golf is a perfect name for the symptom). At work I use Codex CLI with GPT 5.4 - it hits a sweet spot: capable enough that it'll do the right thing most of the time, but unsophisticated enough that I can follow along with what it's doing and cut it off and go, "Nope! That's wrong because x, try y instead." For users like me, the threshold for PMF was never "Do I trust it to run unsupervised" but rather "Is supervising it faster than doing it myself" - and as the models get more sophisticated, they actually go backward on this. Most importantly: a bunch of open weights models hit that sweet spot around the same time - I've been having a go with OpenCode and GLM 5.3 on side projects at home and guess what.
"Rocket, coriander, cumin? Somehow I doubt the Babylonians had Mexican herbs and spices." Turns out I was wrong about where these come from! All three are native to the Mediterranean/Middle East and would certainly have been used in this recipe at the time.
Oh right, yeah I agree. It's the kind of thing I can imagine the harness downloading in the background anyway (or instructing the user to do) at some point so I can see the sense in just skipping that bit. Kind of comforting they just do it the same way the rest of us do - if it ain't broke and so on.
Yes, that's what I mean - maybe that didn't come across cos I said "bloats". I work for a business that builds web apps for corporates - LibreOffice is bundled in most of the repos of ours that I've seen, a bit like Imagick is in the broader sector.
Someone may need to explain to me what is wrong with this sorry. Is it because it bloats the app with functionality that some users will never use? I feel like that is fairly usual for productivity/business apps generally.
> A signal only carries information if it is expensive to fake. The male peacock’s mesmerizing tail of evil eyes works precisely because a sickly peacock cannot grow one.
This is by far the most important insight in this write-up and I will be thinking about it for days to come.
At the place where I was taught this lesson, it was "getting Hasselhoffed" - in short, you find photos of David Hasselhoff in Baywatch (wearing as little as possible, you see? Speedos were very much in in the 80s), and you set it as the desktop. Perfectly harmless, like actual porn would be too far, but it was far enough to be genuinely embarrassing - not least because every Dev gets the same warning on day one (you leave you computer unlocked, you get Hasselhoffed) so everyone knew what it meant anyway. Everyone would, of course, play along - "Didn't realise you liked Hasselhoff that much - bit of an unprofessional image though don't you think?" And so on. If I remember rightly, I literally came in from a cigarette and the first person I saw just went "Do you admire David Hasselhoff's body?" and I already knew what I had done.
I am struck by the resemblance between this logic and the logic behind drinking. I remember back in my 20s one of the devs would come over at the end of the day and go, "Coming to the pub?" The standard answer was, of course, "I might come for one." The word "one" was a kind of punchline all on its own of course, a running joke.
"Machine learning architecture is second nature to us AI researchers, so it's easy to forget that the average person probably only knows the formulas for gradient descent and ReLU."
What an excellent example of designing for the medium instead of trying to make the medium do something it's not good at! Your whole approach shows a real love, like a delicate respectfulness, that I think characterises early games and is less apparent as the systems get better (though I am sure it is still there).
I started building something like this myself (codename HFG - "Heroku for Groq") anticipating what we are indeed now seeing in ChatGPT viz. enshittification. The intention was to aim it at consumers and non-technical users who want a chatbot to help them draft reports, synthesise papers and so on. You've got a bunch of hackers in the comments going "But how can I run my own evals?" I think the answer for them is: this isn't for you.
Don't want to derail what you're trying to do with Echo in case I'm wide of the mark, but yeah even in that case, if you hadn't considered that use case for it, I reckon there will, probably inside six months, be a substantial market for non-technical users who are sick of seeing ads in a service they already pay a subscription for, and who don't care what the underlying model is - or, indeed, don't even understand the concept of an "underlying model" because they interface with AI as a product.
You have already taken the HFG idea way further than I had even thought of yet, and I feel vindicated in seeing someone else do it. I wish you the very best!
Not only do I think you're right, I think put this way it seems obvious - I am almost heaving a sigh of relief having a better historical comparison than "the dot com bubble".
Every point of friction, every struggle documented in minute detail, on a long and bumpy path to some kind of result. The stakes are zero, the whole project an act of pure expression. It is like a kind of oasis in a desert of articles about new ML models. I don't mind that HN is all about new models all the time now, but content like this being on the front page is soothing.
Here in the UK there is an ongoing debate happening more or less behind the scenes around the language that should be used for families of patients nearing end of life. The standard question is: Should we "do everything possible" to keep someone alive? The proposed better question is: Should we "allow natural death"? Any doctor understands intimately that these two questions are equivalent. Understandably, the average person doesn't. Why wouldn't you "do everything possible"? In most of these conversations the argument against just doesn't come up.
"Common sense is nothing more than a deposit of prejudices laid down in the mind before age eighteen." Attributed to Einstein 1948.
I realise this is the Internet, and it's not for me to convince you of anything. But you've formed an opinion on what is normal and "neutral" strong enough to bring it to a thread about something else online. Is that opinion informed, or received?
For example, you say: trans women can compete in sport with cis women, and that is unfair. Is it? I don't see an a priori answer either way. I certainly don't think banning the "promotion" of trans people in public is a viable solution.
Genuinely curious: what makes you think the moral panic around trans people is structurally different to the moral panic around gay people decades ago? All the arguments you've made here are more or less identical to arguments made to keep homosexuality illegal previously.
This is what neutral language sounds like. It sounds to me like you'd prefer to pretend these people don't exist. I'd like to remind you that male homosexuality was illegal in UK as recently as 1967. Section 28 was in force up until 2003. Same sex marriage was illegal until 2014. The ideology you are seeing in this job ad is liberal democracy. The ideology you are defending is something else entirely, and very much not a form of neutrality.
Enjoy? No, but a lot of reading and writing is very functional, for communication purposes for business etc. (documentation is a good example). In these cases the important thing is that it's accurate.