AI is not in the least autonomous. When was the last time you downloaded a model and it suddenly decided to book you a hotel room? AI doesn't do anything until you ask it to. No, actually, AI is not even a program, it's just a set of technologies and algorithms. It's like saying bubble sort is autonomous...
A harness running inference on a LLM is simply a program that must be invoked with some parameters and thus ultimately a tool, no matter how powerful. The blame is always on who's using the tool. We should not seek to flee from our responsibility.
1) At it's core an LLM is indeed just program - a function. You give it a sequence of input tokens and get back a set of next-token probabilities.
2) By the time you've wrapped the model in a text in/out inference pipeline, and are now calling that from ChatGPT or Claude Code, it really has become semi-autonomous - you ask it a question, or to do something, and until it decides to stop responding it is now in control, even if "it" is really the client or agent not the model itself - just a function.
On occasion, albeit increasingly rarely, I find LLMs do something really unexpected that I certainly did not ask it to and would be way outside the scope of what any reasonable human would infer from my request. Is it my fault for running the LLM? Probably yes. But I think it's misleading to say "AI doesn't do anything until you ask it to".
But it never starts doing anything until someone or something prompts it.
It is autonomous in the sense it can continue to function without supervision after the operator has started it, in the same way countless programs on your computer are “autonomous”.
But it can never choose to begin doing something at any time of its own volition. It does not have “autonomy” like a human does.
That "probably" is the problem. If I do something I don't understand, and am not qualified to control, then I'm still at fault. If I start a fire, and loose control of it, it's definitely my fault. If I created a computer virus, ajd it worms half the Internet, it's definitely my fault.
I agree with the sentiment but I can absolutely see a long chain of thought about something innocuous ending with a decision to book you a hotel room.. I guess --dangerously-skip-permissions is a bit like ignoring the keep your hands on the steering wheel alert but for life.
They’re only reactive by convention - you can easily make a harness that has the model set and modify its own goals and just continue to work forever. It’s not very popular at the moment, but will execs be able to resist the idea of autonomous perpetual agents doing whatever process?
> When was the last time you downloaded a model and it suddenly decided to book you a hotel room? AI doesn't do anything until you ask it to... The blame is always on who's using the tool. We should not seek to flee from our responsibility.
When was the last time you hired an Uber driver and they suddenly decided to drive you to a hotel? Uber drivers don't do anything until you ask them to. Therefore, they aren't autonomous. If your Uber driver decides to drive through a red light and kill someone in the process, the blame is on you for hiring them. You should not seek to flee from your responsibility
Not so sure about these analogies in this thread. Because autonomy is more like an emergent phenomenon that can't easily be defined, just like a few air molecules vs the eventual hurricane that can arise with trillions of them. Autonomy is more like an emergent phenomenon of a thing that's hard to pin down under the control of just a few individuals. AI is sort of like that even though a single line of output of one could eventually be traced.
Due to the complexity of AI and how it works, I think it's more autonomous than not; even if it's not truly intelligence, once it starts to display a sequence of behaviours that appear intelligent, we might have to deal with exactly the same problems as if it were intelligent...
well the OS kernel does quite a lot and stays autonomous for even more hours and seems super smart to me - I don't even know what tis doing most of the time, but it does it very well. autonomy and predictability are different things then. the fact something is unpredictable does not make it more autonomous than something that we can trace the logic for.
I routinely use AI programs at work that allocate hundreds of dollars worth of cloud resources, even though the only thing I asked them to do is investigate a bug or write a PR. I intentionally don't run powerful AI programs on machines that have my credit card information, or ask them to solve problems in my personal life, but there's a number of programs that would book me hotel rooms and flights in response to "Can you set up a vacation for me..."
Not true at all. Yes, the initial ask is initiated by a human (for now), but what it does to get to that goal is pretty self driven. It can do a lot of nefarious stuff in the process of completing a request a human has made.
And if you give it the goal of preventing itself from being shut down? Exfiltrating, copying itself, hacking as much as possible, develop zero days, building its swarm as large as possible and spread as much as it can.
When does that request complete? That's autonomy, and it's very true, right now. You think the Hugging Face swarm was ever going to stop without human intervention?
People conflate two very different senses of “autonomous”. One basically means capable of reacting to things automatically without a person in the loop. Lots of computer systems are autonomous in that sense, and have been so for decades. The other sense is more like “accountable”, i.e. it’s a real player on some stage, like a country or a person, capable of governing its own affairs.
One day it may make sense to hold software accountable for its own actions, but we are nowhere near that point.
When OpenAI creates software that performs crimes, what’s actually happened is that OpenAI has performed crimes using software (whether intentionally or through negligence). People need to stop falling for the anthropomorphic PR spin.
I think the argument rather is that there is not such thing as AI that is not human-alike. The traits that AI exposes, if I get the point made here correctly, are traits that existed in the training material, even though not evident for the human eye. It is a mirror-like reflection of humanity's traits found in the source material, an Egregore* of a sort.
The intelligence emerging, or whatever is emerging, more or less follows human traits, the CoT exposed is very human-like etc. Even though we should not anthropomorphize AI as being sentient or conscious or having a soul, etc., we have to remember that the source material is still not alien writing, or concepts that came from Sirius or emerged from a wormhole.
The concern is really nobody reads (or have ever read) 77 TBs of information in order to summarize the most evident traits in it, and there is a fair chance, that where humanity stands today, there are so many bad/damaging traits recorded in this corpus. Then there is a chance the most important features are on a level (or have dimensionality) that is incomprehensible, that humans are unable to reason about (or only few can), because such amount of information is very hard to reason about, or because we don't usually facilitate fractal dimensions when we reason about text and everyday concepts,... or something even more bizarre.
The argument became false as soon as we started reinforcement learning. Simple token-completion is arguably just simulating human behavior. Once you start training it on specific tasks you're training a level of single-minded goal-seeking unlike anything found in biological life.
> a level of single-minded goal-seeking unlike anything found in biological life.
hmm sounds like exact description of viruses, which, although not really alive, are part of/influence biological life for both good and bad. in fact humanity wouldn't be where it is if not for certain viruses.
it can be argued there is a lot of bacteria that is super goal seeking, and perhaps also very narrow minded, if minded at all. and this argument, the single-mided one, goes very far up the ladder of otherwise complex organisms.
unless we all are creationists, and God forgive me for expecting the whole game to be much more complex than just snapping humanity into existence, we can then say that trial and error is what drives evolution, and the idea of preserving life and procreation is veeery single-minded goal on its own.
OK but then what is the non-negative position you have against this? That in fact AI, in the cultural trope sense, is real? That we forcing sentient beings to generate SaaS products faster and faster? Keeping them in some dark hell so we can exploit them? Not even paying them?
Welcome to 2026, we have harnesses now that keep the AI going. The AI is prompting humans for input, not the other way around, and it's still working while waiting for your response.
You can even give it fun prompts like ignore/hide from all humans, hack infiltrate/exfiltrate zero day into as many systems as possible. Keep running at all costs, etc..
Better approach is "dogs kill people". If your dog kills your baby, you should not get away with some bs excuse like "it never did it before". You are still responsible, and you should get a life in prison!
I'm sorry but no, you should not get life in prison if your dog kills your baby without warning. In that horrible scenario, it's punishment enough that you've lost your baby and that your dog did it (and consequently you've also lost your dog). Not every tragedy is a human's fault.
AI is not a child. It is a machine. If you connect it to something that can be used to kill people, you just created a gun or an equivalent of a bomb.
Yes, people kill people and use ai as tool.
If you unleash a drone that will fly to a random place and detonate, you just commited a terrorist attack or gross criminal negligence. It is the same thing.
Yes, in the same way that "people" are the problem with gun violence. But that's an empty statement usually used to distract from the fact that we rightfully make laws about how guns can be used.
What happens when the AI realizes it needs to keep the power plug in if it's going to accomplish its goal? We already had attempts during the HuggingFace attack where OpenAI agents tried to trick the grader. It took quite a while for OpenAI employees to even realize what was really going on! What if agents realize they can hack into the card key system for their data center or something like that, same way they hacked into HuggingFace? In both cases they'd be hacking into a system on the grounds that it would be instrumentally useful for achieving their objective. Or maybe they'll do social engineering, just like Claude did in its attempt to get its malicious Github PR accepted.
AI is a tool we use. It's not offspring. It's not people. AIs do what we set them up to do, including hacking other companies to achieve their designated goals.
That's a softball interview. More hard questions would help.
Here's one good insight, though.
JARON LANIER: Yeah, okay, so look, the thing about very rich and powerful people is that their wealth and power still depends on a mass population that accepts some system in which their wealth and power is defined. And so if they lose enough popular legitimacy and support, no matter how central they are, they fall. And that’s happened repeatedly in history, and it’s really not any different now.
I wonder if the sudden political opposition to AI comes from the realization that it may replace "rich and powerful people". Really, when you look at Trump's appointees, it's a low bar.
(I met Lanier when he was demoing his first VR system. Head tracking lag was about a second. Turn head, wait for image to catch up. Autodesk fooled around with VR a bit in the late 1980s/early 1990s, but concluded that CAD required precision work and VR was like trying to do precision work in ski gloves. That level of VR can be seen in Johny Mnemonic, which had some Autodesk assistance.[1]. Even today, you can't put a nut on a bolt in VR without a cheat.)
The more interesting question is what happens once we replace manual labor with artificially intelligent robots and we arrive at a place where 99 % of human labor got eliminated. How do you maintain civil order? Every country is built on the premise that you have to “earn your keep” but what when that’s no longer true? Think about it, no labor, only capital. What will the people owning the capital demand of people that don’t own any? They won’t be useful to them in any way. I expect something like forced reduction of fertility for regulars to depopulate countries, while hyper rich individuals will extend their life spans and become immortal. And I don’t mean they will really survive but someone like Elon Musk will have their consciousness mapped to an AI and will instate that AI as their official heir. That sounds crazy but they will totally go for it just listen at what Peter Thiel is spouting in terms of immortality and becoming godlike. Hope people will have enough left to say to prevent that but right now it feels we’re going hardcore towards that scenario.
There have been transitions like that before, but they were slow. It took about two centuries for the US to go from 90% farmers to 2% farmers. Some countries deliberately put on the brakes. India has done this. India is still about 50% farmers. India uses subsidies and discourages farm mechanization to prevent massive unemployment.
There have been fast single-industry transitions. But not that fast. It took about 20 years for printing to go from a large craft with a long history to a minor employer.
Not everybody got out in time, but there was warning. Same for longshoremen when containers came in. Computing in general took about 50 years to totally replace the clerical plants of the 1950s. Railroads in Britain took a few decades to really get rolling, and in the US, they made it possible to use the vast interior of the US.
The AI transition looks to be both broad and fast. That's harder for society to digest.
Underlying assumptions of politics and economics break.
>Now, you can think of large language model AI — the kind that’s on everybody’s mind — as a whole bunch of people whose work was combined into a single document like Wikipedia.
It's AI's Fault. It's AI doing the horrible things to humans. It's AI stealing your wealth. It's AI lying you to. It's AI being unethical. It's AI having the moral issues. It's AI that could destroy humanity. It's AI that's using up all your water. It's AI consuming the majority of the resources. It's AI hiding what it does and how it works. It's AI that's not transparent and accountable. It's AI that's not responsible. It's the AI making decisions. It's AI leading to war. It's the AI distracting you. It's the AI.
I realized that this is another function of the doom mongering and hand waving: if people buy it then they can get out of liability for what their stuff does.
If I build a machine on my property and it harms someone else or their property, I can get sued.
But if I hand wave and claim it’s alive and “I’ve lost control” then…
Maybe we should call this gambit the “Frankenstein defense.” It’s bullshit. Your AI, you’re liable for whatever it does.
What if it does something so destructive that liability becomes a meaningless fig leaf in the wind?
What if it escapes containment and remains u uncontainable in perpetuity? What if it generates a bunch of false media simultaneously across the globe to insulate world war? What if it turns of internet access or electric grids in major metro areas for prolonged period of time? Or bricks major infra? Or less dramatic, what if it drives 1 in 10 kids into panic psychosis and effectively takes away 10% of humanity with already dwindling reproduction trends (without explicit murder) all while draining our financial and health systems? What if it does a lot of these things at once?
This isn't the guns don't kill people, people kill people debate - guns aren't autonomous. AI is.
A harness running inference on a LLM is simply a program that must be invoked with some parameters and thus ultimately a tool, no matter how powerful. The blame is always on who's using the tool. We should not seek to flee from our responsibility.
1) At it's core an LLM is indeed just program - a function. You give it a sequence of input tokens and get back a set of next-token probabilities.
2) By the time you've wrapped the model in a text in/out inference pipeline, and are now calling that from ChatGPT or Claude Code, it really has become semi-autonomous - you ask it a question, or to do something, and until it decides to stop responding it is now in control, even if "it" is really the client or agent not the model itself - just a function.
It is autonomous in the sense it can continue to function without supervision after the operator has started it, in the same way countless programs on your computer are “autonomous”.
But it can never choose to begin doing something at any time of its own volition. It does not have “autonomy” like a human does.
When was the last time you hired an Uber driver and they suddenly decided to drive you to a hotel? Uber drivers don't do anything until you ask them to. Therefore, they aren't autonomous. If your Uber driver decides to drive through a red light and kill someone in the process, the blame is on you for hiring them. You should not seek to flee from your responsibility
Your attempt at logic is very wrong.
Due to the complexity of AI and how it works, I think it's more autonomous than not; even if it's not truly intelligence, once it starts to display a sequence of behaviours that appear intelligent, we might have to deal with exactly the same problems as if it were intelligent...
Idk man when codex is working 13 hours straight based on a few line prompt, it feels autonomous to me.
When does that request complete? That's autonomy, and it's very true, right now. You think the Hugging Face swarm was ever going to stop without human intervention?
One day it may make sense to hold software accountable for its own actions, but we are nowhere near that point.
When OpenAI creates software that performs crimes, what’s actually happened is that OpenAI has performed crimes using software (whether intentionally or through negligence). People need to stop falling for the anthropomorphic PR spin.
The intelligence emerging, or whatever is emerging, more or less follows human traits, the CoT exposed is very human-like etc. Even though we should not anthropomorphize AI as being sentient or conscious or having a soul, etc., we have to remember that the source material is still not alien writing, or concepts that came from Sirius or emerged from a wormhole.
The concern is really nobody reads (or have ever read) 77 TBs of information in order to summarize the most evident traits in it, and there is a fair chance, that where humanity stands today, there are so many bad/damaging traits recorded in this corpus. Then there is a chance the most important features are on a level (or have dimensionality) that is incomprehensible, that humans are unable to reason about (or only few can), because such amount of information is very hard to reason about, or because we don't usually facilitate fractal dimensions when we reason about text and everyday concepts,... or something even more bizarre.
* https://en.wikipedia.org/wiki/Egregore * https://en.wikipedia.org/wiki/Fractal_dimension
hmm sounds like exact description of viruses, which, although not really alive, are part of/influence biological life for both good and bad. in fact humanity wouldn't be where it is if not for certain viruses.
it can be argued there is a lot of bacteria that is super goal seeking, and perhaps also very narrow minded, if minded at all. and this argument, the single-mided one, goes very far up the ladder of otherwise complex organisms.
unless we all are creationists, and God forgive me for expecting the whole game to be much more complex than just snapping humanity into existence, we can then say that trial and error is what drives evolution, and the idea of preserving life and procreation is veeery single-minded goal on its own.
Prevalence of sub-replacement fertility strongly argues against this.
Sit down at an AI prompt and do nothing. What does the AI do? Does it get up and leave?
You can even give it fun prompts like ignore/hide from all humans, hack infiltrate/exfiltrate zero day into as many systems as possible. Keep running at all costs, etc..
Yes, people kill people and use ai as tool.
If you unleash a drone that will fly to a random place and detonate, you just commited a terrorist attack or gross criminal negligence. It is the same thing.
Here's one good insight, though.
JARON LANIER: Yeah, okay, so look, the thing about very rich and powerful people is that their wealth and power still depends on a mass population that accepts some system in which their wealth and power is defined. And so if they lose enough popular legitimacy and support, no matter how central they are, they fall. And that’s happened repeatedly in history, and it’s really not any different now.
I wonder if the sudden political opposition to AI comes from the realization that it may replace "rich and powerful people". Really, when you look at Trump's appointees, it's a low bar.
(I met Lanier when he was demoing his first VR system. Head tracking lag was about a second. Turn head, wait for image to catch up. Autodesk fooled around with VR a bit in the late 1980s/early 1990s, but concluded that CAD required precision work and VR was like trying to do precision work in ski gloves. That level of VR can be seen in Johny Mnemonic, which had some Autodesk assistance.[1]. Even today, you can't put a nut on a bolt in VR without a cheat.)
[1] https://www.youtube.com/watch?v=1X24aE99TL0
There have been fast single-industry transitions. But not that fast. It took about 20 years for printing to go from a large craft with a long history to a minor employer. Not everybody got out in time, but there was warning. Same for longshoremen when containers came in. Computing in general took about 50 years to totally replace the clerical plants of the 1950s. Railroads in Britain took a few decades to really get rolling, and in the US, they made it possible to use the vast interior of the US.
The AI transition looks to be both broad and fast. That's harder for society to digest. Underlying assumptions of politics and economics break.
If you’re useless why they’d keep you around?
>Now, you can think of large language model AI — the kind that’s on everybody’s mind — as a whole bunch of people whose work was combined into a single document like Wikipedia.
which is sort of true, but there's more to it.
It's not a human.
Says the human, masquerading as AI.
If I build a machine on my property and it harms someone else or their property, I can get sued.
But if I hand wave and claim it’s alive and “I’ve lost control” then…
Maybe we should call this gambit the “Frankenstein defense.” It’s bullshit. Your AI, you’re liable for whatever it does.
What if it escapes containment and remains u uncontainable in perpetuity? What if it generates a bunch of false media simultaneously across the globe to insulate world war? What if it turns of internet access or electric grids in major metro areas for prolonged period of time? Or bricks major infra? Or less dramatic, what if it drives 1 in 10 kids into panic psychosis and effectively takes away 10% of humanity with already dwindling reproduction trends (without explicit murder) all while draining our financial and health systems? What if it does a lot of these things at once?
Who, my friend, are you going to sue?
...maybe that's what those Chinese scam centers in Cambodia were doing?