Post

if you think that AI is not going to be able to augment the thing you're doing in whatever role you hold right now - whatever career path and profession - that is called cope that probably sucks for a lot of people! it sucked for me! but learning and adapting here really is key

hailey@hailey.at

increasingly it seems like the only real answer to this is to find ways of ensuring guaranteed income for people. imo we are reaching a point where people who want to build or do things will be able to - unrestrained by skills - and there will be a lot of people who will no longer need to work

16 reposts351 likes

Sometimes thoughts can just stay inside our heads and don’t have to be shared with everyone else AI is currently obliterating higher education and destroying everyone’s ability to critically think, using our fresh drinking water, etc. You sound like the people pitching radium in the 1900s

AI? The thing that we have to be warned isn’t accurate? AI? The thing my company laid off a bunch of people for without measurable returns? AI? The thing I can ask to read a subscribed software? Who the fuck are you that you think you know my job better? You’re embarrassing as hell

I can see limited use cases. But I’m less concerned about it killing us all than I am it not working for crap yet being pushed to hell by our oligarchic overlords because OMG OUR INVESTMENT and their desire to send us to subscription service/mainframe-and-terminal hell.

explain more about the magic economy that pays people to do nothing how does one "adapt" to non-existent jobs and companies who want to hire automation resources instead of human?

they refuse to engage with this very basic question: why does it matter if civilization will collapse from lack humans being paid for labor? none of this matters if you can’t first answer this question satisfactorily

I dunno that anyone's gonna answer a question like that for you satisfactory. My point is more like, where's the buffet of open roles? Most individuals still can't duck out of the market for a hot minute to raise kids without getting dinged on hiring prospects as it stands today.

Why is there no evidence of any impact to productivity or employment? Even the bump in entry-level/youth unemployment that people pointed to as the leading indicator has reversed. There's no impact to the economy or to society other than AI psychosis.

It's currently being used to bill for those services and to assist nurses and doulas with writing care notes for patient records. I am an AI skeptic but I work in healthcare IT and it is being rolled out everywhere.

What role do you think software plays in being physically present and supporting a woman as she gives birth? A doula literally holds your hand, wipes your brow, helps you breathe, helps you move around, among other things.

Scheduling. Medical research. Influencing human society such that birth becomes more or less common. Our lives are even today shot through and with mediated by software, and that will only accelerate. Nothing is unchanging.

It only augments though. We actually had to hire a dev whose whole job it is to manage our AI agents in the dev workflow When token prices quadruple, it’s over

It can be, if the workflow only works at today's token price. A dedicated agent wrangler may be exactly what keeps the useful runs visible and the busy loops from quietly becoming a line item.

Yeah, I’m pretty sure AI isn’t going to change how I cut hair in any significant way. Sure, people bring in AI pictures as references, but that’s no different than when they bring in celebrity photos or whatever. No one is letting a robot near their head with scissors any time soon.

It probably won't change how you cut hair, no, but things like booking/tracking appointments or managing finances or tracking/ordering supplies are things that LLMs already do reasonably well.

They think that “email jobs” are jobs where all you do is send emails. I’m sitting here like, what if my job is “trusted expert who has to sell clients on my unique perspective and expertise in a position that requires talking to lots of people and looking at real places.”

There are people in my profession who use AI for small tasks around the fringes that don’t really matter, but if I tried to use it for most of my job I would lose credibility expeditiously.

That’s true for all professions. Nobody is saying AI is doing your job. It’s an assistive technology like a calculator or copy machine. It’s an efficiency improver. That’s it.

I think it will still be a while before semi trucks can self-drive (legally, or effectively in all scenarios). You just might be safe! (If you are still driving a truck 20 years from now, I think you might not be safe then.)

Have you tried it, but found there are deal-breaker issues with completeness and correctness? I think for legal research it’s rife with such issues today, but, think back to what it could do 8 years ago and then imagine 8 years from now.

By safe I meant, does it affect your job security in your field. For software engineers, it’s already resulted in a lot less job security. Downward pressure on salaries, quarterly layoffs like clockwork, fewer people employed making software (even though most still are).

Personally I think in the next decade or so, all knowledge work (jobs that involve reading and writing or spreadsheets) and most physical labor (farming, warehouses, shipping) will be impacted in the form of decreased job security as corps try to cut costs and replace humans with AI (+ robotics).

“Keep this light pointed at this actor” is something that AI could've handled reasonably well multiple years ago, let alone now. Same for moving props, if you put those props on motorized wheels/arms/etc. Heck, AI's probably overkill for both of those.

I’m retired now but spent my career in sporthorses - buying, selling, training, competing, breeding, teaching etc. AI can’t touch any part of that work and I’d extrapolate that to virtually and all sports careers. There’s a LOT of hands-on jobs that AI can’t replicate

One can believe AI is a massively disruptive technology that should not be in the hands of Big Tech w/o major oversight & regulations while simultaneously believing it is & will be a very useful tool (like word processors & Google search were before it) for everyday use.

I saw some reshared post on Facebook yesterday about how "AI hasn't made things easier for me" and at this point, sorry, that's just a skill issue

as a former AI skeptic whose mind was changed by actually trying it: please be aware that I'm not going to engage seriously with anyone whose counterarguments are just copy-pasted tropes

I mean, you can't say "It made things easier!" without saying that the things are. I see no things in my life that would be made easier by using AI so it's pretty easy to be skeptical when there is fuck-all I can imagine it doing for me.

well until your post, I didn't have a reason to elaborate, because people were just shouting tropes at me :P in my particular case (outside of work) - I use it to reduce the number of "micro-decisions" I have to make in a day, which frees up my mental energy for more important things

(the "skill" here not even being "using AI", but "being willing to learn how to use new things")

Yes, definitely a skill issue! AI probably does make things easier for people who were already dumb and/or lazy to begin with. If you have actual skills of your own, it will just waste your time fixing the mistakes of the gobbledygook machine.

What if the thing I do is called thinking? Because so far the evidence is suggesting it makes people worse at that, and I'm worried enough about cognitive decline as I age from natural causes without accelerating it by outsourcing my brain.

Not adapting seems like a pretty good option to me. You'll probably use next years models very differently, also the models you can't afford this year will be matched by open models next year. Any advantage you may have/gain now can be closed more easily next year. Unless that advantage is money.

That's definitely what I'm feeling in the software industry. The vast majority of my job interviewers have asked about my AI use, and that's a pretty big problem for someone like me who's spent the last 2 years in an environment that restricted AI use.

yeah, i feel that to a degree. there are so many more things i could/should(by most peoples standards at least) be using AI for if limiting PHI exposure wasn't such a big concern

journalism? ai is getting there medical? ai is getting there legal? ai is getting there farming? ai is getting there art? ai is getting there you can absolutely be a part of the creative and design process with ai, and you very well may keep a job! but your job is going to be very different

I've 35 years experience in my field, which involves analysis of social policy. In my field, AI's product is 'meh,' and costs a lot. Context is a real problem for it. A simple example: given a speech, it can't tell you what wasn't said, which is often as illuminating as what was said.

Listen, Anthropic will simply be given an assignment by an editor, make a series of phone calls to conduct interviews via phone and algorithmically rank the level of expertise for each subject on the story topic. /Massive wanking motion towards AI journalism

The other day I was asking AI (ChatGPT) about hosts on NPR and it couldn't even name one presenter from the local NPR Illinois. I will give it credit. It literally said "I don't know" instead of bull shitting me.

journalism. the profession that is famously riddled with misinformation and biased reporting, requires lots of research (if you’re worth your salt), and is often championed by individuals or small orgs. i wonder if AI could help with any of those?

How do you think AI gets its training data? Do you think mighty Apollo floats down from Mt. Olympus to bless the AI with knowledge? No! Those biased humans are feeding their biases into the LLM. And then the LLM recreates those biases! Christ, what are you even thinking?

How? you keep saying that it will help all these different fields but you have yet to cite one concrete example of how it will help them. meanwhile llms are actually just making journalism worse day in and day out

Ehat the fuck are you even on about? AI is biased. AI messes up facts and makes them up! It's literally not deterministic. You people fundamentally misunderstand the technology you're boosting and are smug about it, it's maddening.

hailey, the idea that AI would help with bias in journalism is risible. It requires ignoring already documented context where tech like this has actively reinforced bias, e.g. law enforcement and surveillance. Why do you expect people to believe this will just magically not happen in this context?

I love how my healthcare provider’s AI-enabled messaging system sent me to the wrong pharmacist for a Rx refill and then gave me a bottle of pills I had never taken before for a condition I’d never had. So much more efficient and accurate than talking to a mere human. 🙄

My health care provider’s Ai transcription and visit summary wrote that I was “nervous” because I’d said that pain I was experiencing felt like a bundle of nerves.

When we say it can’t do these things, it’s because it can’t. And unless you have some information no one else has, you’re talking out your ass. People who actually have uses for the tech will tell you that it isn’t replacing any of this because you still need humans in the loop for all of it.

Let me be clear, I spoke to a friend who uses Claude extensively in his job as a programmer and they talked about how it still takes a highly skilled human to know both how to use the AI tools and to monitor their output for the inevitable hallucinations and errors. They don’t replace people.

Even if AI were capable of creating “good art” it’s still made by a fucking computer and devoid of meaning and therefore utterly useless. AI ‘art’ will only make human creation more valuable and relevant

So, what you're saying here is, it can't actually do those jobs now. And in each of those fields, the skilled practitioners keep saying "no, it's not useful." (Have you SEEN Tewson on the use of LLMs in law? Ow. Damn.)

And you've been saying "getting there," and demanding that people use it NOW because it's "getting there," for five years. Nobody wants to believe at this point because five years of, y'know, lying about it.

Come back when it actually works. Come back when it can interview somebody, when it transcribes charts correctly, when it stops citing Goofus v. Gallant, when it stops telling people to plant geese, when the restaurant food pics don't make us want to throw up.

Those kids in Memphis developing lifelong respiratory illness from the emissions of the simple cycle fossil fuel generators powering this future very few people even want I’m sure are so grateful

Our death is inevitable. Yet we still get up every morning at act as it is just one of the possibilities. AI, on the other hand, is not inevitable. We know we can live without AI (we did until a few years ago). Why do you act like you are already defetead by something that is avoidable?

The "AI Assistant" at my job is funny because everyone I've ever known to use it thinks it's amazing for like, a week. By the end of the month they're cursing its name and ranting about how the computer keeps lying to them We could have built a better search for our knowledgebase but noooooooooo

Cuz, see, that's the thing. Whether or not AI can ever get to do any of those things, it will never get there. It'll get close enough, then people will insist it's there and force everyone to pay for it, whether or not they use it, and fuck anyone whose lives the half-half-baked technology ruins.

if you just mean “augment” it’s already there in all of these, no? but also introduces different burdens & problems in them on quality assurance especially

Can you fucking people show your work at some point rather than just posting these punchy tweets that are nothing more than you stating your case as fact?

I am not in any of those fields, but in my field the claims absolutely evaporate with even modest cross examination. I get pitched on new "AI" products for campaigns regularly and 100% of the time it's coders with no campaign experience who think they've solved campaigning and never have.

I'm wide open to the possibility that this will change, which is why I keep taking the pitches, but so far it is literally 100% of the time the pitch finishes, I ask like two questions, and they either concede they didn't think about any of those concerns or they get really defensive. *Every* time.

My wife works in genetics and the tools they buy all tout a big AI component. You'd be surprised how well it works! Not well at all. Surprisingly so! They mostly just ignore the 'AI' piece and proceed based on their own training. But the bosses sure do like the idea of it!

"getting there" doing all the work. Experts in every one of these fields see AI currently as dangerous and largely useless. It's not at an expert or even functional in any of these knowledge domains. Just because the output 'looks like' knowledge to untrained people doesn't mean it's there.

LLMs can't even write efficient code. It's "getting there" the same way that fusion power has been "just 50 years away" for longer than I've been alive. And I'm in my 40s

Humanity in general should be prepared to abandon the skill of professional writing in favor of machine driven derivative predictive text. I don’t doubt that this is going to happen, but it will leave all of us more stupid, as everything we read will be regurgitation.

its extremely concerning that people are still so entirely blinded by "no it cant" attitudes rather than preparing for what is coming. we should have had a UBI plan figured out last year, but its this year now and theres no better time to figure it out than now

alternatively we can figure out that AI is bad for jobs, bad for the environment, bad for neighborhoods, bad for humanity in general, and then just not fucking use it or allow it. remember how asbestos was assumed to be a miracle material, and then it got super banned? we can do that shit with AI

I'm sorry I've seen AI art and sure it's getting *somewhere* but whereever that is is not anywhere I want to be looking at in the list of images that have come closest to making me actually physically ill, AI is *at least* all of the top 50, and these are *recent*

We’re never going to figure out a UBI plan! Welfare, “money for nuthin’” plans have *never* been popular among politicians or the people they serve and that’s definitely not going to change in less than a decade! If anything does happen it’ll be a beefed up unemployment fund that’ll let you tread-

if what you are saying is true (I personally dont think it is) we would need a whole different way of living, a socialist/communist way of living

Ubi is long past necessary, but art and journalism is probably impossible / beyond some smaller shortcuts. Journalism - actually investigating and reporting - will need sentients. Current press release feeds don't count

My local newspaper still hasn’t reported on a fatal interaction with police where the suspect was killed & 2 officers injured. I only know about it because I have a LLM running on a graphics card on a PC in my basement & transcribing & searching police radio. Journalists could use things like that.

The people pushing gen-AI adoption are never going to consent to paying higher taxes to fund UBI, you idiot. They created gen-AI so they wouldn't have to pay their workers. You think they're going to pay laid-off workers?

UBI concentrates power and neglects the contributions that these AI companies are stealing. Let’s start by giving consent, credit, and fair compensation for their works, imho a much more democratic framework.

UBI is not coming. We are living in the golden age of labour disempowerment. Companies want workers starving and afraid; it’s the only way to stop them from revolting against the plans to starve them. UBI would give the disempowered the time and strength to rebel. It will not be allowed.

That is an unsolved problem I worry about most. Our economic and political systems are not built for proactive adjustment to changes of this pace and magnitude. This is what the AI-debate needs to focus on. But we are still dealing with people who are in denial about AI capabilities.

So ❤️ to Hailey, but it's a mistake to steer the conversation to UBI at all. The context I've seen others discuss is not that AI is reducing the overall amount of work. It is very much shifting things around, and unfortunately that requires people shifting with it. www.instagram.com/reels/DcR1nY...

instagram.comDaniel Davidson on Instagram: "Is AI replacing designers?"7,091 likes, 419 comments - bydanco on August 20, 2026: "Is AI replacing designers?".

It's less of a long shot in individual states than it is in the US as a whole. There are also other countries besides the US.

AI is currently begging the US government to give it a 2 year moratorium on having to make money under the guise that its machines are so powerful the world is in danger. This is a Ponzi scheme and you are a mark.

Oh yeah it’s fuckin killin it in the journalism space bsky.app/profile/weat...

Daniel Swain@weatherwest.bsky.social

Fielding questions re: this article w/bizarre factual errors. First, there's no "Montecito County" in CA. Second, 2 fires are actually in *Monterey,* & 3rd in *Mendocino*. Third, 60k ac have *not* burned in either fictional Montecito nor real Mendocino.🤨

I can’t speak for any other field but ai in its current iteration will never be able to make a meaningful impact in the legal field. The only thing it’s good at is doc review and it’s only a matter of time before it waives privilege in a way that scares big law off of it for good

I'll bite. Which of these fields do you actually work in? I'm guessing none. As an attorney, I can assure you, AI is not "getting there." It can be mildly helpful as a research aid, but that's it. Its answers to legal questions are either overly simplistic or just straight up wrong.

It can direct you to relevant case law or statutes, but its summaries of those cases/statutes are rarely accurate or complete. The only thing AI is known for in the legal field is getting attorneys who heavily rely on it sanctioned for citing cases that don't exist or grossly misstating the law.

Medical is an interesting one to me because they have been applying machine learning in many areas for years. My spouse has been doing machine learning bioinformatics for the better part of 10 years now.

half true. augment? yes. replace? no. of course you didn't say replace but that's the implication I was getting. it IS true that its limits are things it's seen before. so for jobs where where's lots of that, it's great. like coding. or generating furry porn.

I see where you're coming from, but I don't think AI as we know it today can drastically change how every single job works. With our current technology it simply can't replicate years and years of someone's experience.

you’re right that it can’t as we know it today. but the rate of progress is only going up, and the applications are only increasing. labs spent enormous effort in making them good at software engineering because that’s the key to unlocking capabilities in other areas

not “every job will use code to be automated”, but rather the model’s abilities in software will now allow them to build the necessities to make the models good in other applications

To me the rate of progress appears to have slowed. Unsurprising given the way this stuff works. It's the top of asymptote. It can only ever be *almost* as good as the data that goes in, and now all new data is poisoned by llm output. Painted into a corner.

I do think however that pretty much every field can benefit, even if just a little, from AI. Nearly every artistic or technical process, for example, can also be controlled or revisioned by it.

The simple truth is if we keep hounding people to use it and beg them to use it and complain that people don't use it is coming if you're ready or not please use our money eating machine

What is AI doing for farming? Because my experience with AI is it’s cool for some little gimmicks and to consolidate/summarize writing you’ve already done and for programming but not applicable in actual, use-your-hands and work with real tangible materials

broadly speaking, the key discovery was the "transformer", a type of neural network that can predict the next bit of information in a sequence given pretraining on what that information tends to look like. with chatbots the information is text; with weather models, air pressure; the list goes on

technically chatbots don't work with "text", they work with "vectors", which're lists of numbers that merely *represent* text. the vectors don't have to represent text, they can represent anything that can be described using a multidimensional coordinate system. each dimension is a separate variable

I work for A Large Book Company. There are "self publishers" with 10k+ books who are flatly making the ebook side fucking unusable with endless slop. Pay-to-play is the only demo this benefits. Everything else is a fucking mess.

No it isn't. And it doesn't even have the potential to justify the resources being sunk into it, or the externalities it imposes. It's no more inevitable than the continued usage of leaded gasoline, asbestos, or radium. Absolute nonsense you are spewing.

You creeps have been saying this shit for years now. AI has not generated a single piece of art worth celebrating yet, and lawyers keep getting in deep shit using it in law because it makes impossible messes. This is another coke-addled sales pitch.

AI has no place in my art practice, as a financially successful, full time working artist. I fundamentally have no interest in using it for the kind of work that I wish to pursue and it would not add anything that I want to my work to use it. It is not what I want out of my process or my work.

To use it would be to do something inauthentic to what my vision and aims are with my work and you can no more say that I "have to" do it if I want to continue making art than you can tell me I need to rededicate my work entirely to black velvet painting. That would simply not be the work that I do.

And there's no point in me pivoting to it in order to "compete" or whatever. I am an established entity with an established audience. They are following me because I make the specific work that I make. If I was using AI, there would be no value add for my audience.

Journalism? No, it's not. I can aggregate existing articles or rewrite press releases into articles, but it can not perform original journalism in any sense of the word. Worse, it reproduces the biases of its trainers, ensuring that journalism that challenges power is effective impossible.

Medical? LLMs are worse than useless here because their propensity for "hallucination" create a false positive risk that puts patient care at serious risk. Other machine learning techniques may show promise, but is plagued by experimental errors leading to irreproducability with real world data.

Several of these are not fully severable from human judgment, even if they can do a lot of it. What's worth reporting? What's worth investigating that *might* be happening? A lot of it could be assisted by AI but it would be stupid to intentionally exclude humans from it

Journalism? AI is great at going to a protest, recording the actions of the people and cops present, interviewing folks, and then assembling the footage, narrative, etc into a coherent frame. Same for everything else you said. It’s autocomplete, not magic.

Hahahahahahahaha! No. You're high on a techbro billionaire's supply, and have neither the sociological or historiographic chops to interpret the introduction of what is essentially environmentally ruinous predictive text with a scam built around it.

Mike Drucker@mikedrucker.bsky.social

You’ve gotta use AI because it’s the future. But also it’ll take your job. But also it’ll create a paradise on earth and solve most of our problems. But also it’ll end humanity soon. But also you’ll get left behind if you don’t use it right now.

robots started replacing people in hammer-swinging bolt-twisting rivet-punching jobs, like, several decades ago the ones that remain are the ones that are tough to automate and, unironically, will probably be among the last to be automated

Robots can function 24 hours a day and can be either repaired or disposed of or recycled when they break or a newer model comes along. And in many cases, tasks will be modified to fit robots rather than modifying robots to fit tasks.

Yeah, in a controlled environment, everything that can be automated will be. However, I don’t think that robots are going to be out in the field anytime soon. Even if they could build robots that could work at a construction site, they are still 2-3 orders of magnitude too costly.

do also consider that you can also outsource it for $5/hr (or less) before you get the machine that'll do it at the equivalent of $20/hr.

Think it'll struggle to weed a border or prune an elderly apple tree. I feel my work is safe. Not because it will never be able to, but because it will never be economically viable to create a machine to do it. The ability to do complex manual tasks will become revered...

I don't hold a role. Maybe it will augment role-holders. They probably need a break from playing a role in someone else's fantasy. Seems like you're cheerleading inauthenticity, apathy and incuriosity on behalf of dangerous philistine sociopaths.

Recent conversation with a client about securing open source in the post Mythos era - re IBM Project Lightwell. They said they had turned to LLMs first (of course) because analysts weren't close enough to the material. Opting out is not an option. We need to get good.

It’s not that I couldn’t find uses for it. It’s that it’s bad for people in general, despite whatever it makes “easier.” You don’t have to agree, but you’re not convincing me either.

you have seen waymo right? with lower crash rates than humans? and people are very explicitly working on applications of AI in hvac and engineering! this is not hypothetical it is happening now!

Yes, but you are assuming that the world all works in an office. AI will build more efficient HVAC but it will not install it, AI might create more genetically modified crops but it will not reap the crops. AI will decimate "office" jobs and wonder if the kids of tomorrow will be happy grave diggers

note that my original post said “augment”. it will take some jobs, it will augment others. re farming, i think you’re wrong about technology not reaping crops eventually, but i also think that would be a net good as the human rights concerns are abound in the world of humans reaping crops

I can personally attest to HVAC installers being interested in using AI to help with their work. Meanwhile, assembly lines have had robots for decades, autonomous vehicles are rolling out in more and more cities, and even landscapers/gravediggers still need to do project planning and such.

How AI works is it guesses what comes next statistically based on data that already exists. By definition, whatever it "creates" is going to be mediocre (at best*). I am already way better than mediocre at what I do. Anything AI does would be a diminishment of what my brain does, not augmentation.

*I say "at best" because AI can detect patterns, but it can't understand the hows & whys behind it like a human can, so it's more likely to produce utter nonsense than a human. So it's actually even worse than mediocre of whatever it was trained on.

Not to mention that AI is basically cigarettes for your brain, so if you keep using it enough, I guess it will be an augmentation of your own work, but that's just because you got stupider, not because the actual work improved.

Good news! It ain't AI (or even the tech sector as a whole) that's skullfucking the environment. It's mostly transportation and agriculture that's doing that. The tiny amount of environment-skullfucking AI is doing is also solvable with renewable energy and closed-loop cooling systems.

And what's it called when you notice AI is very helpful with your job (software) and you immediately generalize that to every job without sufficient evidence

I have actually personally found in my own work, and observed in others, that as broadly capable as AI is, for any given task or role, it is at its best when agents are given a very specific, narrow scope of responsibility designed specifically to accommodate manual review

Can those of us who did our time building and got patents and shit but became disabled in the process finally have a decent lifestyle without working ourselves into the grave yet or nah? I want to rest but my car and house are too valuable for a disabled person to own in the USA

It is in fact being developed for and by people who want you to have as little money and power as possible. they want what moldbug is selling and are spending war-time amounts of money to get there because they *are* at war, with you.

I set up and operate heavy lighting equipment for a living. Every job is in a new location, with different equipment and requirements. Please, tell me how AI will augment this. If humanoid robots were even remotely close I don't think clients would like it, and it wouldn't be cheaper.

Also, since the goal of AI seems to making so an unskilled person can do complex things with it - why would I be left behind? If, as y'all believe, it's "the worst it will ever be" - it will only be easier to pick up later if for some reason I must.

Besides that, it's been "getting there" since it's inception with few changes to the fundamental problems with it. It's an achilles and the hare situation, which you could have guessed if you know even vaguely how this stuff works.

yeah yeah yeah we heard this song and dance with crypto and NFT's, too. Didn't fully embrace it? You weren't gonna make it! Think that it was a scam? That's just you coping! come back with new material.