bbor

Oh my GOD -- great catch! Maybe the most Paris Marx thing I've ever heard of. You literally just made my day.

I do wonder how these radical skeptics --Paris Marx, Ed Zitron, Emily Bender, Alex Hanna, Arvind Narayanan & Sayash Kapoor, etc.-- will be viewed by posterity, assuming such a thing exists. When the first catastrophe(s) strike, will they just fade into obscurity, or catch some of the blame?

...on second thought, the most likely answer is that they'll just seamlessly slide into a different brand of "Anti-AI"-ism, neither acknowledging nor facing redicule for the fact that they'd have to completely abandon their core arguments to do so. If I die of a molotov thrown in 2029 from someone in this crowd after spending years begging them to look up, I'm gonna be so miffed.

This is just another sense of "take X seriously" :). In this context we're assessing the truth of political arguments, not assessing whether those arguments relate to consequential political events.

Thanks for the support, RightNutWingJob :) That's a great example of non-linguistic communication!

Hmm, not sure how to respond, sorry — that’s the point of the entire comment. No these systems cannot think every possible thought, but neither can humans. We have no reason to think their cognition is unsustainable in a way that ours isn’t.

The end of your response seems to be about fears of “model collapse” from a few years ago, but I am convinced those are strongly oversold. For one thing, there’s no need to gather new training data ever again; the training data for Opus is clearly sufficient for instilling reason, and that’s a sufficient condition for AGI -> RSI -> ASI.

  “To put me to work that way, without remittance, is a form of slave labor,” said Wolff, professor emeritus of English and winner of the National Medal of Arts.

Wow. Said without a hint of irony, somehow!

Yet again, Anthropic is the good guys. Though I think this is just the same story from a while ago but told through a Stanford lens, AFAICT?

  To me, goals without plans behind them aren't really goals at all

Ok but that's not how specs and plans work -- they are very intentionally two distinct artifacts.

As usual, though, I might just be biased by default towards defending the Python aristocracy; they seem to pretty much always be on the right side of history, even in the little ways!

This is one of my favorite topics recently — a bit of a guilty pleasure. Please share other procedural flora growth if you know of examples!

I’ve found it’s a lot, lot harder than I expected, though I’m targeting aroids (monstera, philodendron, pothos, anthurium, etc) rather than trees, which adds the tricky requirement to get the leaves to unfurl naturally from their petioles. Not quite completely discouraged yet though, sadly!

Ok my original comment was snarky (apologies to all three mods), but you've raised one of my favorite topics so I can't resist engaging a bit more seriously! In the second clause, that is -- the first is trivially true in some senses and trivially false in others, so it doesn't seem very productive to nail down.

So Chomsky thought of himself as a philosopher of science more than anything, and his life mission is/was to answer a fascinating question: what kind of creatures are we? There's been a growing consensus going back to at least Darwin that we're separated from the rest of Animalia by our "reason" (!= intuition != understanding != judgement != volition), but he helped popularize an old theory[1] that puts it kind of beautifully: human language is only possible in beings capable of producing an infinite set of meaningful outputs in response to a finite input.[2] And so far, that list includes only two definite members: Homo sapiens sapiens, and computer programs built upon large language models.[3]

So all that to say: yes, I think you're wrong; the programmatic ensembles that brought us today's breakthroughs are an "infinitely deep well". As long as we have questions to ask, they are not theoretically constrained from answering them.

Of course, this invites some obvious follow-ups on practicality; even the most impassioned ASI boomer wouldn't argue that the exact ensembles used here could answer every question in the universe. But A) that's possible because some infinities are bigger than others, and B) the exact same caveat applies to humans, whose cognitive capabilities are thoroughly determined by accidents of biology, culture, and episteme. The latter is the thrust of Chomsky's famous debate with Foucault, which is absolutely worth a watch for anyone who hasn't seen it (starting around ~18:00, AFAIR): https://youtu.be/3wfNl2L0Gf8?t=1080

---

TL;DR: I ultimately disagree on both points, yes. In the ways that matter, machines are no more constrained than animals.

[1]: https://en.wikipedia.org/wiki/Wilhelm_von_Humboldt

[2]: Many animals are capable of communication, often quite complex and fascinating communication at that; even flora and fungi communicate! But language is qualitatively distinct. You may have heard this in the pithy-yet-mindbending sentiment "we've taught many apes to communicate with sign language, but none of them have ever asked a question."

[3]: It's no surprise in hindsight that trying to build better autocorrect got us to AGI; I find it unspeakably tragic that Chomsky never got the chance to see how thoroughly he's been vindicated, in this regard. He had his stroke mere months into the singularity, and had a quite negative view of the approach before then.

Yeah it’s just one Millenium problem, no big deal. I was gonna crack quasi-Reimann next year anyway, just hadn’t gotten around to it.

All AI is good for is remixing frontier mathematics, there’s no way it could ever make its own. Right?

So what's a "self-driving company"...? The name is just cartoonishly villainous, but by the end it seems like it's just "a company where non-engineers create software"...?

Well the biggest disconnect is that I rarely run the orchestrator model at medium -- I just checked the ArtificialAnalysis index, and Opus-5.5-xhigh is about as much better than Opus-5.5-medium than Opus-5.5-medium was than Opus-5-medium (~5 PP, or ~10%). If I'm letting a model plan changes and then orchestrate their implementation without talking through the details first, it's almost always worth it to me to pay for that extra intelligence.

The second biggest difference, though, is that I haven't used Opus 5.5 in a few days cause I ran out of credits on Thursday! So... you might be doing something right, there ;)

One thing you could check is the files that tend to get read into context, especially their built-in "memory" feature, which can get insanely cluttered if not actively maintained. This is why I prefer keeping everything I can to three other abstractions, none of them harness-specific: reports (roughly per-(long-)turn), policies (human readable, very concise, identified with unique slugs, system-wide), and tasks (Plane/Trello/Linear/etc., mentioned in git commits).

For something like "high level performance choices", a policy to always think about performance explicitly before reporting back could honestly do wonders. As you say yourself, they're typically smart enough to know what to do when reminded of your priorities or asked explicitly. The trick is to not prioritize everything, everywhere, all at once!

...so weird talking about how to best use these agents, BTW. Truly shuffling the deck chairs on our Titanic of a career/civilization.

Woah, my interests! Anyone have any tips or favorite projects to share from procedural vine/flora generation? It's a fascinating space, and this extension seems to do surprisingly convincing job of it

Well first off I love PyPI and am greatly offended on the behalf of thew newly-formed (forming?) council to fix it, but also I don't think that applies in this case. If your server is capable of decoding brotli packets it's at risk -- you don't have to actually use that functionality regularly.

Ok putting aside the "client" thing again assuming I'm just behind the times on python client-server architecture norms: What kind of vulnerability would concern you?

Sure, we can't say for sure that all 4 million of the jobs who downloaded (not built-with, but downloaded to cache anew) this library yesterday are running public-facing servers. But what subset of them might be? How much would it take before it's a concern?

Python feels particularly worrisome to me, as those are the kind of casually-setup servers on local dev boxes for some random little tool that are the most fertile ground for an exploit like this, where creds could be spread around in plaintext and malware could go unnoticed for a long time. And, of course, where the developers responsible probably aren't in the habit of watching the CVE list (yet...).

I hope I'm not coming across as rude -- I am honestly curious!

I mean, I dare say the 4,200,000 Python servers built yesterday alone with this package might disagree...? https://pypistats.org/packages/brotli

I can't find download numbers for extras, but it comes in the pretty dang innocuous-seeming `aiohttp[speedups]`, as well. Also `urllib3[brotli]`. Many, many people will be running these servers without recognizing the name of the library, especially in the agentic era.

I might be missing something obvious in your message, cause I don't see why server/client is relevant. Why would a frontend app speak a protocol that the backend doesn't?

This objectively pertains to this communities interests in a very serious, important way, but I fear it won't last long :( Thus is the truly terrifying part of the rise of the far-right in the west: being in a pro-Nazi party is seen as a valid political stance again. Just another perspective to consider with measured tones and civil discourse.

...ok. Well. Responding politely on the offchance that you mean well and were just in a bit of a rush:

1. You are not necessarily "born with" ADHD. That's certainly a very plausible explanation, but ADHD is a disorder defined solely for diagnostic reasons, using only collections of symptoms; unlike a typical physical disease, no internal mechanisms or underlying causes are posited, on purpose. So saying with confidence that you can never develop it during lifetime is just invalid, putting aside the more philosophical discussions of nature v. nurture and free will that would come up if that roadblock were ever cleared.

2. In the new DSM5-defined paradigm, the three "presentations" of ADHD are assumed to present at different times at different intensities through the typical sufferer's lifespan, with the prototypical progression being Hyperactive-Impulsive or Combined in childhood that leads into Inattentive in adulthood. This is far from absolute, or at least it's supposed to be -- sadly many US psychiatrists still insist on seeing evidence of hyperactivity in childhood before trying out an adult ADHD diagnosis, based mostly on outdated, partial advice from the DSM4.

3. As the flipside of #1, it can also go away without explanation or reason, i.e. go into "remission"; I don't care to gather data on common that is, but I can say with absolute certainty that it cannot be ruled out. That's just how DSM-defined mental health disorders work on a fundamental level.

4. Either way, none of that is relevant to the question of how prevelant it is overall, which is simply a matter of empirical measurement. Without looking it up, I feel confident that it's well below 50%, so objectively "rare" in some sense.

5. I feel like the "some people just get better at managing it" thing is some personal axe you're grinding about people who are virtuous enough to just grit their teeth through psychiatric disorders vs. those who use drugs to cheat. If not: sorry! If so: please don't do that.

Finally, the "edit" button is inline, it just disappears after some amount of time for some odd reason. HN UI never fails to confuse me, as well!

I mean, does it though...? The strongest claim they make is that "timely diagnosis and treatment" (via medication) "may reduce the risk", which seems like a pretty indisputable statement based on the observed corellations.

Anyway, as a side note:

  People on ADHD medicine are also going through mental health treatments and get special care and privileges in society

I mean technically yeah, but the vast, vast majority of psychiatric medication management relationships don't involve therapy. Still more likely to pursue it though, I agree. Is that a bad thing, though? Absent of evidence pharmacological treatment doesn't help, this feels like a time we can just recommend and embrace both.

  the effects of ADHD medication numb their responses and emotions to what seems outwardly stable and normal but deep inside it is an inhibitor of development

Your musings are not at all based in science, but instead personal anxieties. That's fine for you, but please don't casually muse about the secret nature of lifesaving clinical treatments, especially the treatment that is possibly the most stigmatized yet effective treatment in our society (after PrEP, I suppose).

You may have the Hyperactive-Impulsive presentation (or the Combined one), which is relatively rare among adults. I have the other one of the three (Inattentive), and I wouldn't say it raises my impulsivity, though I guess it's hard to know. The meds literally saved my life and career, but I also almost exclusively take them when working or in lowkey social situations so it's hard to say what they might do when, like, rock climbing or whitewater boating or something.

I suppose impulsivity in terms of working on what I should be working on instead of going into some rabbit hole is a type of impulsivity, and that's certainly decreased! I wouldn't call the former state "fun" though... more like "shameful", "irrational", and "distressing". Personally -- YMMV.

Anyway: yes that's plausible, as this review found pretty strong evidence of increased mortality from injury. Still, I'd personally say it's unlikely for two major reasons:

1. Presumably most injury deaths happen to adults, which are relatively quite unlikely to have the HI presentation.

2. There's just so many plausible causes of injuries, especially in a society with cars. The inclusion of "intentional injury" also introduces a factor that's no doubt strongly corellated with socioeconomic status (i.e. being murdered/manslaughtered), another very plausible cause.

Very cool. If the author's here: is this using the new EmbeddingGemma2, presumably? It's been absolutely blowing my mind.

Sure, it’s not a spreadsheet application. But were talking about prose here, not spreadsheets. A lot of that is covered by editor commands and auto formatting, though definitely not the final example.

I think you might just be used to trying to edit markdown without any syntax highlighting…? The Bactic thing is particularly worrying. I don’t really understand how a changed inline fence could reflow whole paragraphs (i.e. effect multiple blocks) in the first place, but even if it were to, it should be obvious as soon as you type it due to changed highlighting.

And search and replace… I definitely don’t relate to that. Regex handles that pretty trivially, but I’ve also never run into the need to replace a phrase that is sometimes partially marked up — seems niche?

I love any appreciation of markdown, but I think this particular essay attacks the very basis of my sphinxian soul in two ways:

1. There’s no acknowledgement of the rich ecosystem of markdown dialects, which add all sorts of incredible features as simple as inline citations and as complex as full roles and directives. They all need highlighting.

2. There’s author claims to like markdown, but then it becomes clear that they don’t like markdown itself, they just like it being behind a rendered WYSIWYG editor. Which… :( headers should look like headers and tables should look like tables, yes, but not at the cost of removing any of the syntax characters! If you do, you’re sacrificing the whole point of the format: ease of modification.

Clearly their heart is in the right place, though. I absolutely loathe sites that don’t support markdown; there is *zero* reason not to do something _this_ important! ~~I dare say it would be something that the three mods and their uninvolved monarch should go to bed every night feeling ashamed of…~~

Yes, many posts in this technology and business forum are about the ongoing singularity. Wouldn't the alternative be very strange?

Ok so first off I haven't done this in the agential era myself. In fact, I've made multiple (small/toy-ish) TUIs in the meantime! So I'm something of a hypocrite off the bat lol -- my only excuse is that I'm pretty darn webpilled, AKA the ultimate cross-platform system ;)

That said: I indeed was referring to rust in particular[1], based on some vague recollections that I finally took some time to confirm:

1. There is indeed a host of relatively mature options for the mapping to the boring ol' native route(s): Qt through `cxx-qt`, FLTK via `fltk-rs`, wxWidgets via `wxdragon`, and Tk via... itself, I guess? GTK also has `gtk4-rs` and `relm` and is supposedly cross platform, but with some confusing caveats. See the very end for the other "just build three apps" options, on that note.

2. The most practical option for time-crunched, not-huge projects (IMHO) is anything that lets you still use web languages for styling; obviously this stretches the sense of "GUI", but it's also wildly popular and is still far better than a full Electron monstrosity. Namely: `tauri`(/`yew`/`leptos`) automatically ties into a separate WebView impl per OS, `dioxus` made their own impl to afford React-like rendering goodies, and `ply` seems like a really promising minimalist option. All actively maintained & growing.

3. It seems the default for not-technically-native-but-feels-native apps (i.e. most apps) is `egui`, which does WASM magic for you across platforms and web (!), or `iced` which seems far too Java-y for my tastes. Neither are at v1.0.0, but I blame the abandonment of semver for that goofiness -- both have sizable teams developing in them full time, with ~30k stars each. `slint` is more mature but seems targeted at IoT/embedded.

4. I'll be honest: I was really just hinting at "whatever Zed uses", since the thread's about text editors! It turns out to be `gpui`, their own custom (yet OSS) engine with a popular external metaframework: https://github.com/longbridge/gpui-kit . As above, does WASM wizardry. More niche options that similarly implement their own languages: `xilem`, `freya`, `floem`, and `ribir` are popular growing options with ~beta vibes (i.e. pre-v1.0.0 but, like, for real this time); `vizia` and `cushy` are more alpha vibes.

5. There's a crowd of "GPU-heavy" (?) frameworks that go beyond my capability to really assess in the first place, but `makepad` is the only one that seems definitely cross-platform. `kas`, `blinc`, `tessera`, `azul`, `pax`, `rosin`, and `rui` (nice snag, that) came up too, but are (pre-)alpha vibes. Game-focused ones also exist, somewhat bafflingly (people make games in Rust??): `imgui-rs`, SDL3 via `yakui`, and Bevy via `bevy_egui` all seem mature.

6. As a bonus there's options to just build a Flutter GUI and attach a rust process to its backside via `flutter_rust_bridge` (genius!) and/or (?) `Rinf`. I've never touched Flutter at all, so uhh take that as you will, I suppose.

7. Finally, to your Swift suggestion: I think you're privileging your OSX and/or Swift preference a lot, there. I'm not sure how that even gets to be cross platform based on what I'm seeing right now, but even if it it's possible, all the OSX-specific apps I see these days (RIP Orion...) put fear into the depths of my heart. What Swift would work for is the megacorp strategy: just implementing three separate apps! Via `UniFFI` or `cacao` for Mac, `windows-rs` for DOS, and any of the open ones above for Linux.

Sorry for a long response -- procrastinating my life for sure, though I have been meaning to do this research for a long time. So thanks for the excuse! Shoutout https://blog.wybxc.cc/blog/rust-gui-survey-2026/ and https://areweguiyet.com/

TL;DR: I'd go with `tauri` for something tame, and `gpui` for a text editor in particular; regardless of what you choose, I think it's a lot, lot more tractable than you expect!

______________________________________________________________________

[1]: After the `bun` debacle turned out shockingly well, I don't think there are "language wars" anymore, just the right tool for the job for the center ~95% of the capabilities/adventerousness bellcurve, so-to-speak. Absent some necessary integration planned in another language and/or extensive experience that you don't want to waste, I think it's just straight up incorrect to pick any "general" systems language (so C, C++, Java?) over rust.

Certainly there are specialized languages that rule in their own arenas --Golang for piping data around, Haskell/LISP/Racket for cool people, Zig for Whatever Zig Does, TS for the web and cloud, Python for science, etc.-- but none of those seem necessary to tie into the GUI.

Oh sure the soviets got a bomb, but I'm talking about the other 95% of the world. The article seems to be about India in particular, which is why it's particularly noteworthy -- India had to pull some straight up Looney Toons shit to build nukes without the US noticing via satellite and shutting them down. And AFAIR they were the last state to get them, other than maybe the time Israel blatantly stole the US's tech.

  When confronted with evidence that the atomic and space ‘races’ of the previous century neither stopped new entrants nor conferred decisive advantages

Seems like a very strange thing to say in a world with so few nuclear weapons. I guess they mean "atomic" in the sense of nuclear power, and the truth that the scientific advantage was slim?

Shockingly all the answers so far are subtly incorrect or extremely unclear.

The actual answer is that this is massive loss for Interpretibilty, AKA "telling whether the model is lying to you or not". Which sounds silly when you're thinking about chatbots, but is an incredibly important line in the sand for superheavy, longrunning, self-improving agents. To say the least!

It's based on this philosophy, FWIW: https://plato.stanford.edu/entries/language-thought/

And a good writeup I just found of recent worries regarding Astra's same shift towards intelligence at the cost of safety: https://www.lesswrong.com/posts/HCRs8btkiamtWSNAL/astra-is-h...

So first off, talking about "LLMs" is kinda a yellow flag that you might be a bit behind the times, here: in ~April of this year[1], "agentic engineering" became truly tractable for serious work. We've all been playing with LLMs for coding and research since ~2023 of course, but they were mostly limited to the chatbot paradigm; even after years of Claude Code "harnesses", most engineers were still not trusting them to perform significant tasks on their own due to inevitable quality issues.

Now, that has changed. This isn't just better models, it's a qualitative improvement in capabilities, a step-change, an inflection point; what the MLrs like to call emergent capabilities, in a loose sense.

To pick just one very tangible example close to my heart: This is why the three.js subreddit[2] has exploded in popularity, showcasing demos by non-engineers that blow any three.js app I saw pre-singularity out of the water, which took hundreds of hours of specialized engineering to make (ask me how I know...). The community is reacting[3] much like the mathematicians, understandably.

---

Ok now to answer your actual question (sorry -- that was the intention, originally!) as concisely as I can. "Software factories" are any system you setup to help you build software in deeply agentic ways. All the major harnesses and "ADEs" like Zed could arguably fit here already, but there's a massive ecosystem of other takes, as this article implies. I don't want to give them too much credit for just snagging the name, but the "Factory" startup does embody the idea pretty well: https://docs.factory.com/

They often (though not necessarily, IMHO) include some services that spawn their own agents after deterministic checks ("latent services", in my lingo) and services that use DL models more subtly to empower otherwise-symbolic services ("semilatent services").

In my biased experience, they often focus on a few central usecases, solving the problems that individual devs run into when trying to run multiple harnesses on multiple projects without friction, longterm decoherence, token wastage, or bottlenecks:

1. Code review. This is the big one and far trickier than it seems (trust me!), so often tried then abandoned. There's a huge startup just for this: https://www.coderabbit.ai

2. Cross-harness syncing of settings, hooks, and/or various flavours of very important markdown -- AGENTS.md files, skills, and subagents personas being the big ones.

3. Cross-agent, harness-agnostic coordination and communication. This just recently reached both the main harnesses, but it seems internal for now; a `BULLETIN.md` helps me a TON.

4. More generally, cross-agent coordination when it comes to worktrees, resource sharing (e.g. OCR queues on my 4090, for me), persisting agent-facing content (e.g. reports, trackers, scratch files, .jsonl blobs) in a convenient place that doesn't clog up the repo, and other technical details that might otherwise have to be figured out again and again by each agent.

5. Enforcing standards, either through hooks (blockers, nudges, or revertable fixes) or the aforementioned services. Think commit messages, prose simplicity, doc conventions, etc., though the sky's the limit, here. Like an intuitively-intelligent, more complexly triggered, agent-only precommit setup, in a way.

6. Finally, there's a whole ecosystem of tooling for agents, which I like to call "ACIs"/"Agent-Computer Interfaces". The sky is both the ground and the limit, here; it's really whatever the dev finds their agents to be oddly inefficient about, or just straight up bad at. I personally some to make edits "self-healing", normalize post-OCR cleaning, and capture quick screenshots w/ optional aesthetic critique[4], if that gives you a sense.

TL;DR: Fun stuff to make that speeds you up because it's tailormade to how you work! I made a bunch because I've been ""self-employed"" and thus had the time to explore myself[5], tho I understand why a more realistic after-work paradigm would make this kind of work wasteful.

On that note, back to writing cover letters -- apologies if the long message bothers anyone. I really should start a blog or get a hobby or something, huh

---

[1]: i.e. Mythos and GPT-6, though obviously access rolled out slower to orgs/people without the special access programs, and even slower to people who prefer open models, where GLM-5.3[-Flash] was the first real contender[9]. For internal agent usage see this graph https://epoch.ai/data-insights/openai-coding-agent-spending and/or this writeup from June https://www.anthropic.com/research/claude-code-expertise

[2]: https://www.reddit.com/r/threejs/top/?t=month

[3]: https://www.linkedin.com/posts/danielgreenheck_i-think-its-t...

[4]: https://gitlab.com/doering-ai/libs/tact , https://gitlab.com/doering-ai/libs/mend , https://gitlab.com/doering-ai/apps/piq

[5]: Most of it is still private, but I made a subset public at https://gitlab.com/doering-ai/libs/fleet , and my code review & hooks are at https://gitlab.com/doering-ai/apps/gate

[9] https://www.anthropic.com/research/glm-5-3-and-the-spread-of...

Ok I'd like to bring up something a bit more bold/controversial: why are we building big TUIs at all? Like... ever? Isn't it pretty trivial to make your rust GUI native across the three major operating systems these days, or am I underestimating that?

I prefer a TUI to a heavy Electron app, don't get me wrong. But Helix (& Mitos!) seem like a baffling choice compared to Sublime Text. Not worse: I literally don't understand the appeal.

I'm just a layperson with strong, strong thoughts about this. So... take it with a grain of salt.

  a U.S. push for “deprescribing”

Always fun to remember that the person in charge of the federal health system isn't a fan of drugs based on vibez. Still waiting to be but in a "camp" to get off my prescribed adderal, as he casually mused in 2025...

  Doctors and researchers know antidepressants don’t work for everyone, and some even agree the drugs are overprescribed.

Kinda absurd framing, even if it's followed by a semi-related "But...". Some scientists doctors and researchers agree that climate change isn't real, that ADHD doesn't exist in adults, and that v-cc---s are a scam cooked up by aliens; you wouldn't so casually make those seem like reasonable positions!

  finding the right treatment is often a frustrating matter of trial and error

That is the very intentional shape of modern psychiatric care, and will be forever until we can gradually decompose each psychiatric disorder into one or more neurological diseases with thoroughly-specified physical characteristics. In other words, it will and should be like this until we have a dramatic scientific revolution in neuroscience.

  One method for doing so, based on brain imaging, is already moving toward approval from the U.S. Food and Drug Administration (FDA) as a tool to guide prescribing.

Presumably they mean EEG, here...? We already have tons of neurocognitive tests that "guide prescribing", but adding a new tech couldn't hurt, especially as EEG & fMRI explode in utility thanks to DL. Still, frustrating that this is framed as an alternative or fundamentally new development, rather than just one more (very expensive) tool in a psychiatrist's repetoire.

   a statistical and AI tool called PETRUSHKA, which they trained on the sociodemographic details and health histories

Hoooooly shit. "Sociodemographic details"?! Did they miss every single controversy around ML in the 2010s, and every public health textbook since then? We've been down this road a dozen times in a dozen contexts; absurd that professionals are trying something so resoundingly harmful.

By the way, the study (linked in article) is shocking. The fact that at least two PhDs signed off on a study with such an obvious glaring issue ("lack of a double-blind design and the large amount of missing data limit the validity of these results") to get a paper out and promote their vibe-coded webapp is a massive indictment of our academic system. Or at least Britain's, I guess. In case it's not obvious: of course people who are told their antidepressants were asigned by a super intelligent machine based on a detailed quiz will be more patient with them than people who get the "usual standard of care", which, again, is very intentionally and openly "guess-and-check" and often opaque to the patient.

  He and his team have trained a machine learning model on patient records from a clinical trial and used it to predict who would do better on CBT or one of two antidepressants.

The next study is also frustrating :(. To cut it very very short: A) CBT vs. medication is not a coherent dichotomy, B) it's repeating very basic mistakes around unintentional discrimination as discussed above, C) the "machine learning model" is just linear regression (!!!), and D) their attempt to cross-validate has serious, serious issues.

  Other researchers are looking for clues to treatment response in the brain activity thought to underlie depression. “That’s really what we’re treating,” says Leanne Williams, a psychiatrist at Stanford University. Her research has identified six distinct “biotypes” of depression based on functional magnetic resonance imaging (fMRI).

Oh jesus christ. They don't teach philosophy of psychiatry at Stanford?? Does she know that she's departing from the sanctioned, licensed framework for all psychiatry? And why is a psychiatrist doing this instead of a neuroscientist, or ideally a specialist in neuropsych?

  The tool might also identify people who will likely never respond to either psychotherapy or first-line antidepressants, Williams says—potentially allowing them to skip straight to second-line therapies such as transcranial brain stimulation, electroconvulsive therapy, or ketamine.

Well that's fucking terrifying. "Whelp your history and race puts you under the curve in our linear regression model, so let's just skip the boring steps in the middle!" - a statement uttered by the deranged.

Sorry for the spam. Just scary and frustrating. I'm at least the author here is trying to frame this in a way that might redirect some of the unscientific, hateful, ignorant urges of the Health Czar.