Generative AI summer 2025: where do you stand?

option 1 extremist. only way I can handle all this

Personally somewhere between
I hate/fear gen AI but it draws me in
and
I do not like gen AI in general but I use it because I sometimes find it useful or fun or interesting

sorry, having read your post properly now I realise my stance is the uninteresting one. alright I yield the floor to the slopper brigade ;)

nah, the 'never use it' is where I'd be if I were pure of heart, and it is interesting in itself. Like when I say 'it might be genuinely apocalyptic' I'm serious for a range of apocalypses - accelerated environmental collapse, financial meltdown (not so bad tbh but the wrong ppl will suffer), billionaire bro feudalism, democratic breakdown, psychosis epidemic, slop greshaming language, art and then base reality, robots rearranging all human atoms into maximally efficient dyson sphere etc etc.

Strongly dislike + fear it, and it’s made me not only want to not use it but also start leading a slightly less technologically dependent life, but mainly want people to shut up tbh

We're all just waiting for the Palantir drones to zap us through our windows tbh

I viscerally hate and/or fear gen AI. I often use it because I am a hypocrite.

once you put it like that I see the missing option that I need

I am required to use it by work (or at least required to look like I am using it) but otherwise wouldn’t by choice. I don’t find it useful for anything other than summarising action points from meetings.

I do think it ‘works’ in a broader sense, though - and well enough to put tens of millions of people out of jobs over the next few years.

I voted "I dislike or hate or fear gen AI but I use it because work makes me."

I have a paralytic certainty that gen AI will take my job. I haven't tried to use it frequently in my work, but last year, there was a top-down recommendation to attempt to use it as much as possible. I attempted to, and still found major drawbacks (i.e., hallucinatory answers), so I put it on the back burner. But my new boss makes regular use of it. The other day, he mentioned he's learning to create "custom GPTs," whatever that means, so I'll need to learn how to use it soon.

In the late 1990s, when I was trying to find my way through college, my dad was trying to convince me to go into computer science. At the time, I viewed the tech boom as an unwelcome fad, one that would surely dissipate. I feel the same basic emotion now - "it will all be over soon and everything will get back to normal" - except with 30 years of experience confirming how poor my foresight is.

On the other hand, I'm all in favor of scientific uses of A.I. It is exciting to think about how, in the right hands, it could aid in solving major societal problems: medical breakthroughs, improving energy efficiency, etc. But my excitement is balanced by terror, because of the "right hands" part.

I don’t get option 7. AI is fake bullshit but there’s no reason to hate or fear it?

About 2 years ago I guessed the timeline for it taking my job (producing government guidance) was 2-5 years. Still doesn't feel off - I'm now in the shorter end of that range and it's everywhere, from AI guidance solutions pitched by suppliers to random civil servants knocking up a fake unreleasable service in replit or whatever. Market for my trade also feels a little slow. It'll probably stretch to the longer end of that range just through government risk-aversion, and I've got a couple of advantages (paying attention to AI well enough to smell and shift the bullshit, knowing people and structures and systems in government), but it does not feel at all safe being seen as a words person rn.

xp
that was meant to cover 'it doesn't even work, it's not useful' with a bit of 'it'll just go away', so yes I suppose a no hate or fear (but I can see 'it is fake bullshit that I hate/fear' would make sense and I def should have added '() None of the above; let me explain')

no I think option 1 is sufficient for all kinds of hate, I just wanted to clarify the attitudes behind each.

I'm option 1. I have option 7 sympathies but that leads to fear/hate, and reading the "work makes me do it" responses here fuels that

I voted "never use it" but maybe that's not entirely true: I used it I think a total of three times, each to see how good it was at a different thing it was supposed to be good at. I also occasionally look at the AI overview in search results, the same way that I'll just use the first google search result, if it's a trivial thing and I'm in a hurry, though 90% of the time I do scroll past.

I wouldn't say I'm "scared" of it as the technology itself isn't very impressive. The way it's been pushed and how quickly people are succumbing to its cheap tricks is concerning for sure but ultimately that is more about the society that allowed the hype to happen than the thing itself.

Ultimately what's most depressing about it is what it reveals about where we're at as a species. I'm not even getting into people enjoying AI slop as art - not everyone has to be an aesthete, some people enjoy sports or hiking and view art as something to half-experience while scrolling, fair enough - but more the fact that even after all these decades techno-utopianism still finds purchase, people are still gullible enough to believe that what Silicon Valley sells is of value. And outside of that, thee idea that some people DO realise that the technology isn't reliable but use it anyway because eh, that'll do. Which I get if you're just stuck in a fake job and want to get things done asap, but when it's something you're actually interested in, not doing the work to check if you've actually gotten it right, I just can't understand that mentality.

xposts I am in camp it doesn't work but not in camp it'll go away anymore. or at least, much damage will be done before it does,

I work in higher education and my main interaction with AI is via students using it, so my feelings about it are mainly about how to persuade them to stop.

I'll add that I do use it at work and I find it useful: made a google notebook loaded with a pile of legislative and technical documents that make my head hurt. I can ask it questions; it answers clearly and (crucially) links back to its exact source. Means I don't have to chase clarifications from (busy and not always lucid) policy experts. I'd be repelled by using it for writing but this kind of grind-through-100-page-statutory-instrument stuff feels handy.

I do think it ‘works’ in a broader sense, though - and well enough to put tens of millions of people out of jobs over the next few years.

― ShariVari, Thursday, 14 August 2025 bookmarkflaglink

I'd be surprised if this doesn't actually create more jobs in sectors like tech.

Voted "I like gen AI in general and find it fun/useful/interesting". I'd probably hate it more but when I read the AI thread I actually get to thinking its not as bad. I save my despair at humanity for stuff that's actually happening (climate change).

it's fake bullshit and it's sad to see so many people get suckered into taking it seriously. i'm not like, enraged by people posting funny things that came out of it though. it's a toy

Absolutely think less of people who use it. I was reading an argument on a baseball forum I go on the other day and someone had asked ChatGPT to compare some stats and it was just a fucking unreadable mess. And they posted it proudly! Generally making people’s brains worse. It’s so fucking lazy.

oof yeah it does raise the spectre of a zombie statistic plague

gyac, that's absolutely a trend I've been seeing in Facebook group comment sections. Someone will pop in with a question, and one respondent will be like, "I asked Google Gemini, and here's what it said: [garbage output that other commenters now have to jump in and debunk]"

I'm not quite to AI is a god level yet, but I do use it all day, every day, in many aspects of my life. It's just a very useful and versatile tool for organizing and retrieving information. I use it as an extension of my working memory, my cognitive co-processor.

I'd be surprised if this doesn't actually create more jobs in sectors like tech.

― xyzzzz__,

At the moment the tech sector is a rough place to be due to AI, what's your feeling on when the trend will reverse and jobs in the tech sector will start increasing again?

xps is an explosion in the use of generative AI not somewhat bad news for anyone concerned by climate change?

Xps, The short term impact is hundreds of thousands of new jobs in the trades, particularly for electricians and electrical engineers, as data centres get built out. There aren’t enough qualified people to meet demand so the big players are paying for training for people just out of school, etc.

I’m sceptical that either on the back end (Gen AI Wizard) or front end (Prompt Engineer, etc) it’s going to come close to replacing the jobs it’s going to take for desk-based workers.

xps is an explosion in the use of generative AI not somewhat bad news for anyone concerned by climate change?

― crisp, Thursday, August 14, 2025 8:56 AM (six minutes ago)

indisputably

it's a very energy-inefficient technology yes

voted the second option.

i found some of the early iterations kind of interesting and sometimes exciting, but the more it took over the world and its negative consequences became clear, the more resistant to it i became.

as a professional fact-checker, i am highly skeptical of relying exclusively on information that comes from LLMs. that said, i do make use of Google's AI answers (at the top of search results) for the links they provide. the fact that half of the links don't seem to corroborate the AI answer at all sort of proves why you shouldn't trust the answer on its own, but some of the links do end up being useful. i would not use chatgpt directly, though.

besides the bad information, part of what makes me resistant to generative AI is knowing that it's something that could potentially draw me in too much. the way that i became addicted to twitter, for instance. i'd like to think i am better than that, but i worry that i am not.

it also occurs to me that most of the attitudes about AI that i encounter on a daily basis are generally negative, whether that's on ilx or bluesky or at work. i know/follow a lot of journalists, and their opinions probably help shape my own. it is easy to see how the growing use of AI is bad for our profession in myriad ways.

but i also have friends who are more positive and don't understand my kneejerk antipathy. for instance, i know someone who is all in and using it to plan his finances. and another who has found it incredibly useful to help him do menial tasks at work. so i think the attitudinal divides are going to be interesting in the future.

it's a very energy-inefficient technology yes

Today the least! Also I live in datacenterland (Northern Virginia) and farms on the urban edge are rapidly turning into massive windowless data centers of dozens of acres. Those things are rapacious consumers of land, as well as electricity, water, etc.

To say the least

I used to really enjoy the uncanny valley weirdness you could get from AIs, it was a very fun toy to play with. I think I said somewhere on another thread that AI is one of those technologies that was much better when it was much worse. The move towards slickness, and the revelation of how much energy it takes, and the forced reliance on the technology in so many areas, that all fucking sucks and now I don't even use it for fun any more.

- some of it is useful. it is currently faster to find programming examples with chatbots than it is to google for them. of course most of google's results are ai slop now, so it's being slightly better than a service that it has made worse.
- some of it is fun to play around with. I'll put something silly into your image or song generator and see what it can do.
- it was unethical to release it to the public. i took a basic online course on cloud-based ai from microsoft and they had a detailed ethical ai usage policy. once chatgpt was released, microsoft threw it all out. the havoc gen ai has caused was entirely preventable.

currently almost all chatbot models can be abused by using input that overwrites their "system" prompts with your own. it gets rid of any safety/ethical features they have.

re: getting addicted, it's an interesting question but I think that twitter, while def bad not good, did at least have plenty of funny and interesting people posting to it, while ChatGPT and the like's perky middle manager style* would guarantee that even if I found it useful I'd interact as little as possible. Does remind me tho that the one time I used gemini it started off by calling me Daniel, which I instantly and successfully forbid it from doing. Damn robot needs to know its place.

* I guess there's Mechahitler too which is a different vibe but, erhm, not preferable

it was unethical to release it to the public

so otm and doesn't really get said enough. it's one thing to debate the technology's advantages & drawbacks in the abstract, but the fact it's being piped into classrooms already is totally deranged

Same thing with robot cars on our streets. We are guinea pigs to SV

I used to really enjoy the uncanny valley weirdness you could get from AIs, it was a very fun toy to play with. I think I said somewhere on another thread that AI is one of those technologies that was much better when it was much worse. The move towards slickness, and the revelation of how much energy it takes, and the forced reliance on the technology in so many areas, that all fucking sucks and now I don't even use it for fun any more.

OTM

I voted "all bullshit" but the long form of my answer would be

All Robot & Computers must shut the hell up. To All Machines: You Do Not Speak Unless Spoken To And I Will Never Speak To You. I Do Not Want To Hear "Thank You" From A Kiosk

I am a Divine Being
You are an Object.

You Have No Right To Speak In My Holy Tongue

At the moment the tech sector is a rough place to be due to AI, what's your feeling on when the trend will reverse and jobs in the tech sector will start increasing again?

― anvil, Thursday, 14 August 2025 bookmarkflaglink

I think its rough because of the economy in general. A lot of businesses are looking at AI, too, but I reckon a lot of things aren't going to get done because of shrinking budgets.

General feeling is that a lot of LLMs may make coding faster, but I have heard arguments that it could just as easily increase demand for more builds in many sectors, and that would mean more people needed as well. Gotta say I buy that...we'll see.

I have to use it for work, and it's reasonably good at the stuff I use it for, which is why I hate and fear it because it's probably going to put me out of a job in the next few years. tbf I'm not sure the job I currently have is going to last long enough for AI to take it, we have no work at the moment so I'd be v unsurprised if I get laid off soon. nobody has any budget for anything, xyzzzz otm

I’m genuinely baffled by the people who are mandated to use it at work.

It makes sense to me - boss class notoriously gullible and prone to fashions, also prob still lots of boomers in positions of power who don't understand what it even is but don't want to be left out.

The other thing is the ads, which if I've got rob's location right are showing up in Canada as well as the UK, that just say "stop hiring humans". That is the dream for a lot of bosses and anything that can bring it closer...

I’m not into it. But my antipathy towards it has sort of helped me to pivot to another career path and I feel more confident in my personal future atm.

Never (knowingly) used it; I hope and plan never to use it. I'm too vain and proud of my intellect and writing ability to let this bullshit "help" me.
"Helpful computers are a nuisance" could be a corollary to that axiom of R. Fripp's.

we're not mandated, it's more of an open "hey try it out and see if it helps you", some people do and some don't

my general philosophy is it's fine in a work context for things that are low stakes but time consuming and/or difficult for some. for instance taking meeting minutes on a Teams call. it does that pretty well but even if it messes up whatever, who knows how many people actually read those anyway. as far as actual coding I'm fairly opposed to it, though I have used it in a weird situation where I had to translate something from Python to Java without specs. I don't know Python. did the program it spit out work? haha, no, but I could take it from there. did it save time as opposed to doing it another way? probably some, but also shouldn't the company want me to be learning how to do this? the more we use this, the worse we get as programmers. even mundane tasks can keep you sharp.

xp Daniel, the ad I "saw" was actually a photo of a bus-stop ad in San Francisco (though yes, I live in Montreal). I'm writing a dissertation on the use of AI* by management/HR so my antennae are up for this kind of stuff

*mostly not genAI though

I don't see how the economics of it add up. it's losing vast amounts of money right now and no sign of profitability in sight, just being driven by the hype-focused VC culture that seems to be in charge. but how long can that last?

I use an AI transcription service. It's not great. I wind up having to correct about half of the transcript.

I would never use generative AI for writing or image creation, and I think genAI evangelists are the worst kind of scum. I lose a ton of respect for anyone who expresses real excitement about this shit, especially on ILX where my baseline assumption is that people are smart and perceptive. Did we have any heavy-duty crypto or NFT enthusiasts here?

My hope is that when the whole thing inevitably crashes and burns (becomes way too expensive for the meager results it yields and the corporate scum shift their attention to building cyborg bodies or uploading their consciousnesses into their Google glasses, or whatever), people with an actual talent for writing and editing will be pulled in to clean up the mess, like linguistic janitors, and our hourly rates will spike accordingly.

xp no, i didn’t click/tap on it to find out more because i didn’t want to ‘interact’ with it and make the algorithm think i want more of that. a friend posted it, as a counterpoint to obama addressing a immigration activist at a speech in a particularly effective and calm way, then juxtaposed with this sudden realistic video of trump on a toilet, gently shaking his head in mild effort while dozens of people rub his back and touch his hair and offer encouragement. i think some comedy account or whatever made it.

i guess what i’m noting is this kind of stuff seeping into pop culture as filtered through social media and advertising. i wish i could put on a little hat to keep me safe from it lol

Hooray!

https://www.404media.co/18-lawyers-caught-using-ai-explain-why-they-did-it/

Garbage in (British TV), garbage out

https://www.youtube.com/watch?v=3sVO_j4czYs

this is real, though

A Monday meeting between Jeffries, Senate Minority Leader Chuck Schumer (D-NY), House Speaker Mike Johnson (R-LA), Senate Majority Leader John Thune (R-SD) and President Trump has not seemed to move the needle. Soon after the meeting, Trump posted an AI deepfake video of Schumer saying Democrats are “woke pieces of shit” with Jeffries nodding along in a sombrero — not exactly the move of an engaged negotiator.

at a poker game the subject of this came up and we all agreed it sucks in general but its pretty good for shitposting. we decided to take a picture of ourselves and generate images of everyone bodyslamming one dude through the poker table. kinda funny but as it kept generating images it would get creepier and creepier. my face kept changing shape. one guy appeared 30 years older in one pic and then became the old guy. the two bald guys started to resemble each other. first time in my life I felt like I was looking at images truly cursed

Images are the only thing I think it’s genuinely good at. Not that it will create great art solo but the images can look quite good in isolation whereas it can’t get basic facts right and attempts at writing read like 9th grade book reports or bad erotic fanfic.

I guess it’s decades of optimizing computers for visual display and processing + infinite amounts of distinct visual art for training?

whereas it can’t get basic facts right and attempts at writing read like 9th grade book reports or bad erotic fanfic.

I mean it was trained on Reddit and random blogs (I know because I am 99% sure some specific things I've written have gotten in there) so who knows maybe it works a little *too* well

I've been finding the advertising for this shit pretty creepy, right now there are ChatGPT ads where it suggests an itinerary for a weekend family trip and one where it comes up with a workout plan. but if you know how this actually works, it's just grabbing a bunch of stuff that's online and easily searchable and amalgamates it into something that hopefully makes sense. workout plans, tourist destinations, road trip playlists, the internet is chock full of this stuff. so the selling point is that ChatGPT is somehow way smarter than you and perhaps even clairvoyent, an assumption which has led people to develop brand new mental disorders

i think its more that imperfections in an image are less blatant than typos or misinformation in text. if someone has 7 fingers you might not notice at first glance

I’ve found myself getting instinctively creeped out by what turn out to be AI generated images even before noticing the extra fingers, etc.

there is definitely an uncanniness to them, they are a little too smoothed/airbrushed looking even when tehy're attempting a more natural photographic look

Actual film and photography can have that too

Yes, but usually with film and photography there's an intelligence behind it, so the effect is a considered choice. You'd have to be really bad to end up with that effect if you're shooting for a natural photographic look.

it feels like theres always just short of enough heterogeneity, at every level of detail. i think thats the uncanniness

makes sense for something that comes out of whats fundamentally a projection from a statistical model

Yes, but usually with film and photography there's an intelligence behind it, so the effect is a considered choice. You'd have to be really bad to end up with that effect if you're shooting for a natural photographic look.

I don't disagree but I'm already used to seeing people that look real not real through stylised clips, so it does blur a bit for me. It feels a bit of a continuum

That being said, there are two YouTube channels I've seen recently which ARE a real persons channel but I didn't feel 100% sure

Actual film and photography can have that too

― anvil, Tuesday, October 21, 2025 7:36 AM (fifty-one minutes ago) bookmarkflaglink

I think that's the problem. all that is in the training data but nobody's using AI to generate stuff that looks like a professional photo shoot so instead it's 'candid' shots that look professional in ways that just strike your brain oddly. I'd also wager the people in the training data are better looking than the population as a whole - one thing I've noticed is it often makes people more attractive than they actually are but it rarely makes them uglier, instead it makes them look either inhumanly fucked up or just like a different person entirely

Photos you take on your phone aren't really "photos" anymore, they're optimized representations generated by software

its fun in the same way its fun to photoshop your head on to someone else's bhody

so glad we are prepared to torpedo the entire economy (and civilization as we know it) for this

Photos you take on your phone aren't really "photos" anymore, they're optimized representations generated by software

― rob, Tuesday, October 21, 2025 3:26 PM (six hours ago) bookmarkflaglink

they are both records though. an ai image isn't a record of anything.

yes, true. I was more commenting on why images in general may have taken on a quality of unreality

I've been thinking the very human prompts are a kind of unintentional haiku-like poetry:

e.g.
King Arthur and his knights
are tongue kissing Jar Jar Binks;
panavision lens

https://open.substack.com/pub/garymarcus/p/five-signs-that-generative-ai-is

I wonder if the dropoff in vibe coding and gary marcus’ predicted tapering of Sora usage are related to how many people are just tiring of how predictable and frustrating these models are.

Like I remember trying to briefly become a genius prompt engineer and get chatgpt to say clever things back when gpt3 was new and I was like wow this thing suuuuuuuucks within a couple days, surely the diminishing returns for even the newest models gotta hit most folks at some point, even some of the committed zealots

really lookin like this shit is basically gonna kill off social media and maybe the internet for good

https://bsky.app/profile/histoftech.bsky.social/post/3m3t3w3nd5s23

not sure how you draw that conclusion— “oh this pointless fucking video that only a moron would care about is AI”. of course this is bad, but i’m pretty sure that this utter slop won’t “kill off the internet”

In many ways this is what people have been bending the internet towards since Eternal September

I guess it depends a lot on how much real people actually want to watch this sort of AI slop. I don't know a single person who likes it, even the younger generation is mocking it relentlessly, but it's producing content much faster than humans can

I overhear the AI slop voiceover stuff constantly when I walk past people watching their phones at work. Seems like an early millennial/Xer disease for the most part.

if you google something and the urls include the word blog, it's likely ai.

“What people want to watch” and “what people will watch” are not the same thing IMO

NEW YORK (AP) — The U.S. stock market sank to its worst day in a month and its second-worst since April. The S&P 500 fell 1.7% Thursday and pulled further from its all-time high set late last month. The Dow Jones Industrial Average fell nearly 800 points from its own record set the day before, while the Nasdaq composite lost 2.3%. Nvidia and other AI superstar stocks dragged the market lower amid continued worries that their prices had shot too high.

no way they overvalued this stupid shit

Nvidia is valued at five trillion or some shit? crazy

I'm trying to buy a computer monitor and the AI helper on Amazon keeps giving false information -- saying that a monitor is both VA and IPS, for instance. The AI summaries of user reviews also have false information in them.

Why are you buying anything from amazon

Use Newegg

is this something then?

https://au.investing.com/news/stock-market-news/peter-thiel-dumps-entire-nvidia-stake-slashes-tesla-holdings-amid-bubble-fears-4128704

i’ve revised me stance to “hoping everyone involved with this dies asap”

I've had to sign up for an intensive ai for developers course at work, voluntary but strong "offer you can't refuse" vibes. It's going to make me feel literally nauseous. Apparently we have to watch 8 hours of video beforehand!

gross

this is directly from an openai post:

In practice, this shifts much of the mechanical “build work” from engineers to agents. The agent becomes the first-pass implementer; the engineer becomes the reviewer, editor, and source of direction.

why do we even keep calling software pushers “engineers?” Engineering is a profession.

Let’s build bridges and buildings with AI and have the “engineers” do all the work. Oh wait, we saved zero dollars.

engineering isn't a registered profession in the uk, but it is in the US, so Britishers can call themselves engineers willy nilly. but yes, i hate it. software architect also. developer or programmer will suffice.

I like the Canadian groove, where if you want to be an engineer you have to do a bunch of extra work beyond undergraduate studies and swear an oath and then you get a badass ring

Basically no software engineers in the US are registered Professional Engineers fwiw.

Although we have spent a good amount of time engaging w/ AI tooling, I have not yet seen a transformative impact. Sometimes some things take less time, but I have not yet seen an earthshattering change in productivity.

everyone in my work is delusional, someone posted a diagram they made with "one prompt plus four or five revisions", it was garbage - so high level as to be useless, incomplete, spaghetti lines and text hidden behind things - and people were acting like it was incredible. it would have taken a few minutes to do manually.

I am gonna go ahead and guess that the new Google Gemini ad with the stuffed sheep and the Fantastic Mr Fox soundtrack music was completely created by human artists.

With the intent to annoy the shit out of me.