The glasses are great! Considering they have a light that shows when you're recording and it is NOT easy to disable, I definitely don't think they are 'pervert' glasses or spy glasses.
There are certain recordings where glasses make the perfect form factor.
I think most people would not continue to code if their coding agents stopped working, just like if the internet went down.
I agree its not on the same level as the internet/electricity but its been quite a lot more impactful than you are giving it credit for, and that has only been increasing over time.
Totally disagree. Newer models are doing much more advanced things in domains like math, programming, and cybersecurity.
I don't see why a company charging more for less is a surprise or would lead to thinking that AI models have reached their limits. Thats an incredible claim and it would need incredible evidence. Astra is much more capable than Luna. Luna being 'good enough' for most of my daily work just shows how strong Luna is, not that Astra isn't much more capable.
As models get more and more capable, they will only appear to be at the extremes. You'll need to see them solving more Millenium problems to notice a difference as the tide is continuously raised for all models as well.
Some frontier labs are reporting positive "adjusted EBITDA" (earnings before interest, taxes, depreciation, and amortization, with extra adjustments to make the figure positive).
Free cash flow (operating profit less investment), actual cash coming in, is deeply in the red.
EBITDA can be a sensible measure of profitability when there isn't much need for additional investment. That doesn't seem to be the case with these operators. They need to invest aggressively to avoid losing customers to competitors. All of these operators have made multi-year commitments to invest more in infrastructure. In addition, they have guaranteed quite a bit of debt to fund it.
Maybe it all will work out fine (and I sure hope it does!), but I didn't see any hard data from the OP, or from you, supporting that view.
EBITDA might make sense for the resellers who package up open weight models and sell inference. It is not appropriate for the labs who have billions in debt for RAM, new data centers, gobbling up competitors, etc.
Those real debt obligations are going to want to be paid back.
Definitely. The question is: Is it enough to recoup the enormous capital costs and justify the level of investment they've received. I think there's a decent chance that it will be. But maybe not. And the longer they keep focusing on training new models more so than on inference, the more uncertain I become that it's all going to work out.
Honestly at this point with training costs I don't see how it could ever pay itself back unless you get RSI, in which the talk of money really isn't the main problem any longer.
We're in a situation where AI isn't going to go away, but whatever financial mode we're in right not is not going to work.
Labs are playing money games with EBITDA, which is not uncommon, but also hides the extent to which they are in the red (deeply, deeply, in the red, and projected by them to get worse).
> AIDE demonstrates that recursive self-improvement at the harness layer can produce transferable gains in an AI research agent’s research efficiency. During the recursive self-improvement run, the loop accepted seven rewrites, each under a fixed evaluation budget (section 3.2). Under this fixed evaluation budget, gains in optimization capability on AI R&D tasks translate to gains in research efficiency."
Go talk to an artist, a junior developer, or a musician. And then realize that the goal is to generalize the damage to every other intellectual domain.
This same argument is easily dismissed by most because it boils down to "why not dig with spoons?"
It only becomes compelling to me if you believe in true AGI/ASI, at which point all human labor becomes irrelevant, and then we have something quite new to think about.
But simply "it automates away a small portion of jobs" is not compelling.
The industrial revolution wiped out like 90% of human labor as it was imagined at the time, but it nearly fully replaced it with new forms of employment.
There are a lot of unknowns with what this would look like with AI.
Don’t forget the effect on students! It has the potential to be a great learning tool, but nothing is in place for that to happen. It more likely just kills all reinforcement learning
I have the opposite experience. My intellectual curiosity has increased immensely since starting to use Chat GPT. I can snap a picture and instantly know what kind of bug I am looking at, whether I am looking at a Red-Tailed hawk or a Red-Shouldered hawk and how to tell the difference, what the gauges on my water tank and boiler mean, and numerous other things that I am curious about daily.
These are things around me that I wouldn't have bothered to investigate just due to time and effort. And now I appreciate all these little curiosities more than ever.
>These are things around me that I wouldn't have bothered to investigate just due to time and effort. And now I appreciate all these little curiosities more than ever.
I think you're conflating access to information with genuine learning.
Being able to look something up real quick isn’t the same as knowing something, is it, especially in the specific absence of some reinforcement mechanism to retain it?
I just mean that I’ve been able to casually look things up once in a moment of curiosity on the internet since the 90s. This did not lead to memorising all those interesting facts, much less understanding them, and I’m not sure what the mechanism is that would lead to that sort of knowledge acquisition. If anything I’ve probably learned less because I know I can just look something up I’d have otherwise needed to learn. And in any case, actually learning hasn’t resulted in a collection of random otherwise unrelated facts I can remember, but understanding systems and relationships.
The knowledge I have learned has indeed come from what might arguably be described as, maybe not as painstaking, but certainly something with intention and effort.
To me it sounds like you’re describing learning to write by having ChatGPT write essays for you. You don’t actually know how to identify birds, you know how to ask your phone to do it for you. Though I can understand the illusion this may cause you to feel. I get the same when I watch high level chess players play, and then am reminded what I can actually do my next game.
For me, the primary mechanism for learning is doing projects.
And, doing projects in domains I'm unfamiliar with (and those I am) is massively sped up by AI.
I learn to do these projects with AI - it's different than learning to do it without AI, and I don't learn some steps I otherwise would, because AI can do those.
But, I start and finish and use and polish so many more projects in so many more domains, with AI, than I could without.
And, starting and finishing projects is the point for me, not learning obscure stuff that doesn't help there, so, I don't feel like I've missed out on much. You might look down on it, but, if I have AI and you don't, I will bet on me to finish a project before you, better than you, in almost any field where we have similar levels of experience.
IDK about frontier-pushing-pursuits, though. I've written algorithms by hand that I doubt AI ever would have, in its current state... but, then, at least a few of those I'd given up on, for lack of time, and when AI came along, I was able to go back and have it finish them. But, for me, the work wasn't the goal, or even my own accomplishment therein - I just wanted the result. If your motivations differ, you might not find AI as fucking awesome as I do.
Sure, you can be much more productive writing essays for example if you have AI write them. Homework of course is much faster and easier if ChatGPT does it for you. All of these things were possible even pre-AI by paying someone else to do things for you (though in some cases it would be considered cheating, but if we set that aside). You’ll definitely blow anyone out of the water who studies or writes for themselves. I also totally get that the productivity gains are so attractive you don’t mind not learning some or all of it. But we are talking about knowing and learning things. Or at least that’s what I thought the thread was about.
"If I should really WANT to answer the foolish question you have just asked, or any of the other questions you have been asking me, let me remind you that I have a row of electric push-buttons on my desk, and by pushing the right button, I can summon to my aid men who can answer ANY question I desire to ask concerning the business to which I am devoting most of my efforts. Now, will you kindly tell me, WHY I should clutter up my mind with general knowledge, for the purpose of being able to answer questions, when I have men about me who can supply any knowledge I require?"
- Henry Ford
This is roughly my point. The moment the AI was able to do things for me, those things which it could do became trivia. I never had a taste for trivia. I still fill my days with learning and knowledge... the things I must learn and know to do my work and run my business well. I am growing faster with AI than I could without - it's allowed me to expand my purview while narrowing my focus. No less learning, just a different shape of it.
So, to me, the argument that "AI doesn't help you learn" completely misses the point. AI can help you learn, if you use it to and apply yourself, but also, AI can allow you to accomplish without filling your mind with "general knowledge" - not the high value stuff that is your business, but the low level stuff that was incidental.
So tell me about the red tailed hawk. Why is it interesting, and how will you apply that in your life?
According to current theories of learning, the best learning happens at the point of struggle, where you internalize that you went down a wrong path, and needed to backtrack. What part of the struggle of learning did you experience as a result of this, and how did it reshape other thought patterns?
I don't know what to tell you, it's just a tool man. Like a Google search or a trip to the library.
Why don't you tell me about your struggle to research and discover all the science and technology behind indoor plumbing so you don't have to take a shit in the yard, or in your pants?
I mean for me personally I have always had an obsession for knowing these things anyways, it just lowered the barrier to entry for me a little bit in terms of information gathering
First I went from asking my parents then reading lots of books then specifically researching on google and forums now for most general knowledge information I can just ask AI
I'd like to argue the other way around, it can be bad for some people while still being extinction-level bad for entire species. I'm not talking about sentient AI going Matrix, i'm talking about the real material impact of datacenters being built so that a small minority can enjoy funny tools.
Oh, they're more than a drop in the ocean for sure. Which doesn't mean there's not other things to address, where population is definitely not the main problematic factor, but consumerism is (and throwaway culture). This kind of whataboutism is fallacious, you can do better.
EDIT: As a small anecdotal point, any new human activity that is measured in GW of power consumption is not a drop in the ocean. Admitted, some of these data centers projects are "only" in the 100s MW scale, but all of this combined is increasing electricity consumption by a vast amount overall.
I looked (*asked AI), and AI data centers are projected to hit 850-920 TWh by 2027, if you count the energy required to produce all of the buildings and equipment. This is about 3% of global electricity consumption, or 0.5% of global energy (counting fuels, etc., outside of electricity).
So, yeah, that's a lot.
Also, there are at least a few data centers popping up around me (11 GW Fermi, > 1 GW Crusoe). I live in a desert, and water is scarce, and what's used is mostly non-renewable (pumped from an aquifer), and these data centers evaporate water at an alarming rate.
So... yeah, that's pretty significant... much more so locally than globally, but globally, too.
Global electricity consumption is up about 600% since 1970. So, that 2027 projection of 3% of global in 2027 is equivalent to about 20% of all electricity consumed in 1970. A fair amount, for sure.
IDK if it's worth 3% of global electricity consumption. I know I'd be happy to spend that percentage of my own home's electricity on it, for what it does for me, and I think the world would be better for it. The alternative is accomplishing far less... easily halving my productive output (10x is my honest estimation of productivity boost from AI). I do not believe in the value of my labor, I believe in the value of the product thereof. So, IDK... hopefully it's not a zero sum game, and AI (or motivated people with AI) can help it all pan out.
It's not just the data centers, but it's part of a bigger trend of massive industrial projects depleting natural resources for questionable benefits for humanity. There's basically two approaches to ecological problems: either pretend some wonder technology (AI?) is going to save us all and keep driving straight into the wall, or seriously question any industrial activity in a cost/benefit analysis.
Our society is currently headed in the first scenario, building ever more useless shit and accelerating environmental collapse, pretending CO2 capture, AI, unlimited energy storage, agro-bots, desalination and on-site recycling of complex materials will magically keep life sustainable despite any scientific evidence showing otherwise.
There are some projects I'd consider to be useless, but it doesn't seem very likely that data centers will be useless. Maybe not used for their original purpose, though?
That's a good point, but i doubt we will find useful applications that require resources on that scale. The only usecase i can think of is more mass surveillance, which in my view is a net negative.
> ChatGPT was shown to support and even encourage dangerous drug use, with statements such as "Hell yes—let's go full trippy mode" and advice on reducing his Xanax tolerance so that a single tablet will "fuck you up"
> It has literally contributed to suicide and murders
So have many pharmaceuticals that are generally regarded as beneficial. Suicide is a side effect of more than a few, but doctors disregard that, believing that the results are worth the risk.
Honestly, a lot of people could stand to be a lot more informed about what they're getting into. People could use a lot more skepticism when interacting with AI, social media, news feeds, and basically every other form of media, especially those originating at the top of some pyramid or another.
I don't tend to like controls or regulations, though. Not that they aren't needed, but I don't trust the people responsible for writing and enforcing them. More often than not, they are just the top of the pyramid squashing everything beneath it.
How proportional is it to deaths caused by other technologies by usage?
With large enough numbers the existence of any extremely rare event becomes certain. Consequently claiming there are instances carries no information value.
Proportionality is the only sensible measure. How much more dangerous are Chatbots compared to, say, teapots?
That’s close to whataboutism.. just because other technology can also be dangerous doesn’t absolve OpenAI/ChatGPT.
Also, ChatGPT is clearly exerting a massive (potential) influence on people if their MAU numbers are to be believed, so rightfully it should get attention on the potential harm.
Do you think all the comments in here are positive about the model? Because they aren't. In no way does it seem astroturfed. And yeah, pretty boring to read that kind of comment every time.
Engagement is generally more important than purely positive sentiment.
Anyway these comments were made when this thread was in an earlier state. I agree that it has gone on to be more "organic" looking. That doesn't exclude it initially being manipulated to the top, in my mind, but certainly they aren't carpet-bombing with only booster comments.
There are certain recordings where glasses make the perfect form factor.
reply