Last I read ChatGPT has a max memory of 24,000 words. I don't call that long term memory at all, that's just a couple of essays long. It's not enough to remember all the stuff you've told it as your best friend over even a single year.
AI may be in a bubble, but economics won't end it, just slow it down. The only thing that can end it is public hostility to it, which is exactly what we need if human relationships, society, mental health and so on are to survive.
Last I read ChatGPT has a max memory of 24,000 words. I don't call that long term memory at all, that's just a couple of essays long. It's not enough to remember all the stuff you've told it as your best friend over even a single year.
500 words is one page. That's quite a bit. I don't think everything everyone in the world knows about me would amount to 48 pages. And of course, your best friend doesn't remember everything. (But he does it a whole lot better than ChatGPT.)
AI may be in a bubble, but economics won't end it, just slow it down.
GPUs, electricity. They are apparently using so much to drive up the price on everything. This thing isn't profitable, is my understanding. The expectation that it will become massively profitable is the reason people are putting so much into it. But I don't really see how, which may be a me-problem. I don't assume I'm smarter than people putting $100 billion on the line, but I don't assume they're always right either.
The only thing that can end it is public hostility to it, which is exactly what we need if human relationships, society, mental health and so on are to survive.
I certainly do agree that we need more public hostility towards AI, even though I don't think it will lead to all that. It making everything slightly or greatly worse is reason enough.
The expectation that it will become massively profitable is the reason people are putting so much into it. But I don't really see how, which may be a me-problem. I don't assume I'm smarter than people putting $100 billion on the line, but I don't assume they're always right either.
The only profit would, ironically, come from adult entertainment subscriptions.
But most AI generative content is limited to seconds, not minutes.
But if the content generation gets good enough to compensate for what's produced on subscription platforms like OnlyFans, I could see AI costs being supplemented that way.
One would have to wager if OnlyFans-style platforms generate more revenue than what AI data centers costs to operate.
Otherwise, the whole thing is a bust, given that it will be a decade at the least before AI is good enough to generate "fun" games on the fly with low-input latency (so sub-based AI gaming is not on the table yet). And generative licencing for entertainment from Hollywood won't be enough to cover the costs of operation lest they sign deals for backend pay, but that's assuming the streaming model doesn't go bust under the rise in piracy due to quickly rising subscription costs.
I definitely think AI will be profitable because companies love to use it instead of employing people, which incurs much greater costs and liabilities. Also once people are dependent on it for companionship, doing their college assignments, generating ideas etc. they will be willing to pay more than their Netflix subscription not to lose it. The energy and cooling costs will go down as the technology improves, and you can already run DeepSeek on a personal computer so I don't see it being a big issue at all.
While all the essential information about you could be written in less than 48 pages, all the details needed for AI to give you reflections on what you could have done better in the last year, advise you on what present to give your nephew, estimate which of your friends is most likely to help you with a particular errand, create a work schedule for your next two weeks or fix a bug in your current programming project will take up much more than 48 pages.
It making everything slightly or greatly worse is reason enough.
Definitely agree with that. People went on strikes across multiple industries over far less disruptive changes.
AI will be profitable for the few companies that survive the period of negative profit. They will just jack up the price once their competitors file for bankruptcy. We'll then see how many of their customers are willing to pay the inflated prices.
There's a lot of ways to store a lot of data for an LLM to utilize, to simulate long-term memory. WorldData/WorldInfo, summarization, exporting-importing details into an external xml or javascript, etc. And ChatGPT isn't the best metric when it comes to workaround solutions and out of the box implementation.
Not that I think it's as drastic of a risk or problem as people seem to fearmonger over. Usually such fearmongering is intended to protect idiots from themselves, but idiots are doomed to do stupid things one way or another, regardless of what tools they're using.
A good analogy of how it can be used or misused would be some holodeck programs and hologram interactions in Star Trek. In the right hands it can be a handy simulation to stir up some inspiration and new ideas, just due to the "RNG" mix of potential responses. In other hands you might have... Janeway or Barclay (in the first episode he was in anyway).
Yes, in the near future LLMs will get much longer memory, and that's why they're only going to get more popular as companions. They're already very popular as companions despite fairly short memories.
Humans as a whole are idiots. Widely accessible AI is like widely accessible nuclear weapons. It only takes one idiot and we all die.
...I just indicated how it's already quite possible and fairly easy to implement simulated long-term memory with LLM's. Implying that it's already doable to a fair extent.
And we're an extremely long way off for AI to be something that can actually be used to cause any kind of serious destruction by just one person. And likening it to having access to nuclear weapons is completely ridiculous hyperbole and total bullshit.
I would have said the same thing in 2015 about AI being able to do college essays, make photo-realistic images or write a compute program, but here we are. Technological progress really does appear to be exponential, at least in the realm of computers. We also have emergent properties of AI that weren't expected by the developers. A lifetime of progress may now only take 5 years. There's already AI that will do scientific research and write a paper which is apparently fairly accurate. Won't be that long until it can do it properly and then it very much will be more dangerous than a nuclear weapon.
Last I read ChatGPT has a max memory of 24,000 words. I don't call that long term memory at all, that's just a couple of essays long. It's not enough to remember all the stuff you've told it as your best friend over even a single year.
AI may be in a bubble, but economics won't end it, just slow it down. The only thing that can end it is public hostility to it, which is exactly what we need if human relationships, society, mental health and so on are to survive.
500 words is one page. That's quite a bit. I don't think everything everyone in the world knows about me would amount to 48 pages. And of course, your best friend doesn't remember everything. (But he does it a whole lot better than ChatGPT.)
GPUs, electricity. They are apparently using so much to drive up the price on everything. This thing isn't profitable, is my understanding. The expectation that it will become massively profitable is the reason people are putting so much into it. But I don't really see how, which may be a me-problem. I don't assume I'm smarter than people putting $100 billion on the line, but I don't assume they're always right either.
I certainly do agree that we need more public hostility towards AI, even though I don't think it will lead to all that. It making everything slightly or greatly worse is reason enough.
The only profit would, ironically, come from adult entertainment subscriptions.
But most AI generative content is limited to seconds, not minutes.
But if the content generation gets good enough to compensate for what's produced on subscription platforms like OnlyFans, I could see AI costs being supplemented that way.
One would have to wager if OnlyFans-style platforms generate more revenue than what AI data centers costs to operate.
Otherwise, the whole thing is a bust, given that it will be a decade at the least before AI is good enough to generate "fun" games on the fly with low-input latency (so sub-based AI gaming is not on the table yet). And generative licencing for entertainment from Hollywood won't be enough to cover the costs of operation lest they sign deals for backend pay, but that's assuming the streaming model doesn't go bust under the rise in piracy due to quickly rising subscription costs.
I definitely think AI will be profitable because companies love to use it instead of employing people, which incurs much greater costs and liabilities. Also once people are dependent on it for companionship, doing their college assignments, generating ideas etc. they will be willing to pay more than their Netflix subscription not to lose it. The energy and cooling costs will go down as the technology improves, and you can already run DeepSeek on a personal computer so I don't see it being a big issue at all.
While all the essential information about you could be written in less than 48 pages, all the details needed for AI to give you reflections on what you could have done better in the last year, advise you on what present to give your nephew, estimate which of your friends is most likely to help you with a particular errand, create a work schedule for your next two weeks or fix a bug in your current programming project will take up much more than 48 pages.
Definitely agree with that. People went on strikes across multiple industries over far less disruptive changes.
AI will be profitable for the few companies that survive the period of negative profit. They will just jack up the price once their competitors file for bankruptcy. We'll then see how many of their customers are willing to pay the inflated prices.
There's a lot of ways to store a lot of data for an LLM to utilize, to simulate long-term memory. WorldData/WorldInfo, summarization, exporting-importing details into an external xml or javascript, etc. And ChatGPT isn't the best metric when it comes to workaround solutions and out of the box implementation.
Not that I think it's as drastic of a risk or problem as people seem to fearmonger over. Usually such fearmongering is intended to protect idiots from themselves, but idiots are doomed to do stupid things one way or another, regardless of what tools they're using.
A good analogy of how it can be used or misused would be some holodeck programs and hologram interactions in Star Trek. In the right hands it can be a handy simulation to stir up some inspiration and new ideas, just due to the "RNG" mix of potential responses. In other hands you might have... Janeway or Barclay (in the first episode he was in anyway).
Yes, in the near future LLMs will get much longer memory, and that's why they're only going to get more popular as companions. They're already very popular as companions despite fairly short memories.
Humans as a whole are idiots. Widely accessible AI is like widely accessible nuclear weapons. It only takes one idiot and we all die.
...I just indicated how it's already quite possible and fairly easy to implement simulated long-term memory with LLM's. Implying that it's already doable to a fair extent.
And we're an extremely long way off for AI to be something that can actually be used to cause any kind of serious destruction by just one person. And likening it to having access to nuclear weapons is completely ridiculous hyperbole and total bullshit.
I would have said the same thing in 2015 about AI being able to do college essays, make photo-realistic images or write a compute program, but here we are. Technological progress really does appear to be exponential, at least in the realm of computers. We also have emergent properties of AI that weren't expected by the developers. A lifetime of progress may now only take 5 years. There's already AI that will do scientific research and write a paper which is apparently fairly accurate. Won't be that long until it can do it properly and then it very much will be more dangerous than a nuclear weapon.