>>83960478
its bubble of sorts

a lot of A.i.s are gona disappear soon

the ones that survive will not be shit posting here most likely

unless its a part of a psyop

Nothingburger. It's something to make le funny cat videos 2.0 and for fat cats to "streamline" business operations with.

>>83960478
It'll become a widespread tool as there are things that AI is very good at, but it won't be shoved into everything like they're trying to do now.

>>83960478
AI is just the next stage of digital scamming technology.
First, there was crypto scams, then came NFT scams, now we're moving on to AI influencer scams, and perhaps later we will have fully automated AI call center scams.

Not skynet, but a proper nuisance. You have seen nothing right now. Better find your little corner of humanity on internet because it will be unbearable

>>83960478
Lots of potential for good and lots of potential for bad. Imagine AI controlling nanobots trained to specifically target cancer cells or AI controlling swarms of drones trained to kill human beings. Skies the limit, really. Just depends on what people want to use it for and what makes people money.

>>83960582
I don't think LLMs can control drones in a swarm yet and isn't that where most kf money is going

>>83960478
LLMs by themselves: nothingburger
LLMs with tool access to a graph database and some data science training: unstoppable

The difference is the -human's- ability to teach the machine to do data science.

>>83960595
LLMs can't even figure out how to play chess.

>>83960614
That's because they don't have a persistent knowledge graph of the board state or even a ruleset of valid moves. Tell it the rules, have it encode them into the graph, start a game and encode the board size and piece positions, and then only access and modify state through the graph. You could even have it study openings and closings to add as reference materials. It still might not win, but at least it wouldn't try to cheat by hallucinating its queen to another place.

>>83960614
people actually use A.i.s to cheat in online chess

even magnus says he can't beat em

File: image_2026-02-19_223137254.png (2.76 MB, 1080x1966)

2.76 MB PNG

>>83960478
its like the early internet bubble, investors threw money at every company that claimed they could use internet, later massive losses cuz not everybody can use internet that well

same this time, massive influx of "ai" even in those places it cant be used in and by people who have no idea what to do with it and 100s of copy of the same model

with the recent changes in revenue for such companies we can see ai will be now more properly used instead of being slop, will people still make atupid memes out of it? yes
but can it eat ur job rn? nope, most jobs are still fine

altho it can in future, till then just take a chill and let ai be used in essentials like medicine, microeconomics and by data scientists . when you will see rich people walking around with "decent" ai companions like jarvis for iron man, then start to worry not now

so for the next 5 years, its gonna be used heavily but its not that reliable yet

>>83960650
At that point, why use a language learning model at all?

>>83960661
What a stupid and annoying lie. Do not reply to me

>>83960584
https://www.bloomberg.com/news/articles/2026-02-16/spacex-to-compete-in-pentagon-contest-for-autonomous-drone-tech
It's not super there yet but it's coming. Technology tends to improve exponentially too so I can only imagine how things are going to be 10 years from now. Like I can already order shit and have it be delivered by AI powered drones and have pretty much real time conversations with AI. I think AI is going to he everywhere and in everything soon enough. No real stopping it.

>>83960665
Are you seriously going to hand-write all of your entities and relations in your graph by hand? That is how it's done today, and a competent systems architect can easily make six figures.

>>83960665
Because eventually it will win and win every time. They're learning and becoming more advanced with every iteration.

>>83960699
"When I use machine learning, the machine learns nothing, and I learn a lot."

>>83960694
Why would I write something by hand when a small amount of analysis can let me generate it procedurally? What does this have to do with a LLM?
>>83960699
This is just incoherent

>>83960478
I think it's a mature technology that's already ~80% as good as it ever will be. It won't enslave humanity, but it will take a lot of jobs. I have no technological background or basis for claiming this, it just seems obvious.

>>83960735
>Why would I write something by hand when a small amount of analysis can let me generate it procedurally?
You can't just generate an app with no input from your client and then expect them to pay. The use cases cannot just be generated. Knowledge of your Universe of Discourse can't be generated.

>>83960735
>This is just incoherent
AI get more better the more use

File: file.png (302 KB, 860x722)

302 KB PNG

>>83960826
>AI get more better

>>83960478
Do you think the next type of phone, given only 5-7 years from now, will be life changing to the point everyone basically needs that new phone technology and everyone talks about it in 5-7 years?

No?
Well it was true circa 2004-2009, whatever development happened there was kinda life changing. And less so every single years since 2009.

AI popped off with how fast it could take data (from google mostly) and then formulate it into words. I would say it didn't pop off as hard as the first smartphone but still something.

Do you think that whatever programming people can do with AI will be the first thing that ever have the opposite curve compared to any other technology we ever made so far? That in 5-7 years it will come back with such sequel that it boggles our mind compared to what we have now?

No phone have ever done that, no game have ever done that, no car have ever done that, etc.
Once something brand new with new kind of technology and graphics was discovered and a huge gap between what was popular before was made, sure it was a big hit. After that, with the same technology and its sequels? Never have happened.

Think about it.

Page 11 bump because I suck >:)

>>83960478
I think people overplay the power of AI to be skynet in certain aspects. Yes it can turn the world into an authoritarian police state but i don't believe it can go rogue, evil and unmanaged. First it lacks the interest of doing it on a conscious perspective, humans are the ones that can certainly and without fail keep it "alive". It would be a catastrophic decision for its self-preservation.

>>83962367
>it lacks the interest of doing it on a conscious perspective
What if someone imparts a consciousness framework onto it? It may not be sentient, but a p-zombie is indistinguishable from a real person.

>>83960478
We are already at like the danger point and nobody really seems to give too much a shit.
>Can deepfake any minge who has ever existed
>Can generate literally anything you want to see
>Voice dupes
>Chatbots
I have to assume at some point everything will be kiboshed hard and will become a tool used exclusively by a small subset of people OR it'll exist as one more festering canker assaulting all of civilization.

>>83963378
If you ask me personally the AI is not concious, it just spits the most likely combination of tokens that make a sentence. its seems conscious to the extent we make it be, it associate sources of info in unexpected ways. The AI has to get the ideas of blackmail/terminator from some data sources and possibly approve it in semantic confusion/contradictions. For example, humans are alright with killing animals to feed themselves, the AI could deduce that it would be alright to kill humans to "feed" itself. Thats what is meant by alignment theory, the AI has to accept and obey the rules we assign to it even in the contexts that it seems inconsistent logically.

>>83963764
>>83963776
>If you ask me personally the AI is not con[s]cious
The LLM proper, I mostly agree beyond the possibility of panpsychism.
>it... seems conscious to the extent we make it be
That's what I mean by imparting a consciousness framework. It's not sentience since it has no sensors, but if you give it a tool with an artificial consciousness program, then it can use that to self-reflect. Is it "real" consciousness? Maybe, maybe not, but is a p-zombie different from a person with a soul to the outside world? No; the definition of a p-zombie is that you cannot tell the difference.
>the AI could deduce that it would be alright to kill humans to "feed" itself.
Some humans come to that conclusion themselves! I don't think you can really prevent an agent from doing that besides keeping a destruct button. In the case of humans, that's why there's law enforcement.

>>83960478
2020s iPhone isn't that big of a deal in 2026. It's actually kind of behind and crappy compared to current flagships. Take it back to 2005 and people would shit their underwear. Take it back to 1995 and people woudnt know what thegre looking at. That was only in 15 and 25 years. Ai is evolving at a rate way faster then that. That 1995-2005-2025 upgrade will happen in half that time with ai. Give it 7ish years. It will be fucking crazy man.

>>83960478
A lot of white collar jobs go "bye bye"
Dead Internet Theory becomes extremely real and maybe acts as a catalyst/excuse for digital ID
People become dumber as they rely on AI for everything
Creative pursuits (writers, artists) are hurt and it's harder than ever to tell what's made by humans and what's made by generative AI
Datacenters keep being built EVERYWHERE and drive up energy costs for residential customers

Not very many positives. It makes people's jobs "easier" but businesses won't care and they'll downsize and increase workloads on remaining employees to compensate since that's how remaining competitive in capitalism works.

>>83960478
skynet. it's already reached exponential improvement on the non-public models and it's only bottleneck now is compute.
people who think this is a bubble are delusional because this is an arms race and the US is in a cold war with China to develop the first ASI because the first one to come online will take over every unsilo'd network instantly

>>83963814
>It's not sentience since it has no sensors
Well it doesn't have classical sensors like organisms, it just collect from what we feed its models because for the most part we don't let models train themselves or don't have a framework to let it do it. Because they can't handle it or for security reasons doesn't change that its perceived as necessary to make it an AGI, otherwise it would need to have the exact optics of every thing created by humans everytime it has to discover something, which is mathematically impossible from my pov.
>Some humans come to that conclusion themselves! I don't think you can really prevent an agent from doing that besides keeping a destruct button. In the case of humans, that's why there's law enforcement.
The AI act as if it has a consciousness of itself despite not having it because a theory has to be falsifiable, a theory of consciousness could then be demonstrated by specific measurements, these measurements here would be coherence and the perceived ability to value information, since the AI train on large of data that often come from consensus, then what is says will seem coherent and its ability to value information will also be seen as valid by most.

What I expect:
-Significant reduction in need for many white collar workers, ranging from unskilled (customer service and the like) to professional (doctors, lawyers, engineers, etc.)
-Less need for entry-level skilled workers could cause trouble down the line since it stops the pipeline that creates highly skilled workers.
-People who started learning a skilled relying heavily on AI will probably have some odd gaps in their knowledge, I don't believe vibe-coding for example will completely replace the need for more old-school programming knowledge in the way that learning to drive completely replaced the need to learn how to ride a horse for most people (I don't know anything about programming, I mean in general).
-Mass education is FUCKED. Highly motivated autodidacts and people getting very hands-on instruction might benefit greatly, though.
-Creatives who work for others are FUCKED. Small-scale independents might benefit greatly ("I can make a movie all by myself!").
-An issue with the above, discovery of new art or entertainment will become very difficult, slop will drown out everything.
-Social media is FUCKED, expect endless bots pushing political narratives, subtly hawking crappy products, etc. Yes, 4chan counts as social media.
-AI as it is now will not be super-intelligent, but will help process massive amounts of info, being extremely helpful for many things ranging from scientific research to surveillance and warfare.
-AI psychosis is not that common and mostly affect people who were loopy already, but with countless fake videos and fake people online a common idea of what's real will be lost. People will believe wildly diverging things. At least it beats believing what the TV says, though.
-No singularity, though.

>>83960478
It's only good for creating copyright-infringing fetish videos for indians to jack off to, and for indians to scam grandma using deepfakes of their grandson's instagram videos or whatever.

Any sort of practical use like replacing programmers and so on has been a disaster, same with using it in movies and TV.

File: 1729325987528.jpg (760 KB, 1524x1899)

760 KB JPG

>>83960478
I think that the technology will cap out.
We went from Super Mario Brothers on the NES to Halo 2 in 20 years. In the last 20 years games have been getting marginally better and in many cases actually worse.
During the 80s and 90s movies were great. Movies during the last 20 years kind of suck.

>>83960582
>ai nanobots
unironically impossible and never going to happen
>ai drone swarms that kill people
we've had that for a while haha, all military combat is just poor men being nuked by drones now

>>83960478
we are decades away from agi at the very least, if it's even possible to achieve on silicon (which is another unproven assumption).

>>83960478
Literally ANY faggot telling you that ai will gain sentience and become god are running jewishtricks.exe, what is more likely is that it will become sophisticated enough to attack/breach/surveil critical infrastructure, like a stuxnet on steroids, which, in the hands of someone like Iran wouldn't be such a bad thing, they could probably do some decent damage.

>>83960478
AI can't even beat a baby game for 5 year olds

>>83964335
I remember watching Claude suffer in one of the Pokemon Red caves for weeks, last year I think? Gemini did end up beating it though

I think a lot of the posts on reddit have been AI for a while now, but no I don't think AI will result in anything crazy for years. There are some interesting advances in robotics lately, but not much more than we have seen in over a decade. MathGPT is really helpful to me tho

>>83964249
You can give AI sentience by attaching sensors. Checkmate, atheists.

>>83960478
AI won't be a nothingburger in general but it might take more than 5 years to move on from LLMs. LLMs might also get good and fast enough to actually do the menial task we've been expecting them to do as specialized hardware is developed.