> From what I know, the optimizations of LLM's don't really apply to robotics, so all this funding of billions going into LLM only to pivot into robotics is a bit :/ for the investors.

AI wave is more than just LLMs. Movement autonomy for cars/robots, image/video generation, protein folding, etc. Those are not LLM based AI applications. They are all downstream from the transformer architecture. Autonomy AI development is the missing piece of robotics, which is why so many billions are being invested now.

The lack of moat regarding LLMs is a problem only to those playing in that field, but their actual goals are not just running LLMs, they are like I said aiming for actual general intelligence.

In the mean time, companies are training or optimizing their own models for their use cases, like the ones I listed in the previous reply. They do have a moat, because they require specialized knowledge to play in that field. Even the abel police guys, their competitors were just an interface to ChatGPT and it worked abysmally.

> IMO When I mentioned S&P AI stocks, that's exactly things like Google,microsoft,amazon which are still similar to OpenAi and anthropic, don't you think?

Absolutely not. OpenAI is a private company spending insane billions for a moonshot project. The public S&P500 companies are investing their insane profits and making a return on those investments. Their infrastructure and scale is a moat.

▲

Imustaskforhelp a day ago | parent [-]

I had written a original draft of an reply but this conversation is getting really interesting and wanted to write it again lol.

I agree with all the aspects of protein folding / general purpose automation due to "AI" and not LLMs but that was happening before the "AI" hype thanks to OpenAI / chatgpt where so much money was flowing into it...

And I have no issues with them if their prices were baked into realism that they were baked into pre 2022's / whenever chatgpt got launched

My biggest issue which is the crust of this discussion might be that I believe that tech stock prices are roaring so high mostly because of the AI hype that they bring which raises their prices.

Oracle made larry Ellison the richest person for some time due to the stargate project / due to their deal with OpenAI / then larry invested into openai for some hundred billion $ which openai paid back to oracle and oracle's stock price increased more... rinse and repeat?

The thing is, why is oracle which I think is S&P company raising because of their deals in LLM's at an unastronomical rate/ unprecedented rates.

Google/meta/microsoft/amazon are also all integrating AI into their every project / mentiniong AI as much as possible which lets be honest again, is mostly LLM's for the most part.

Yes I know, google has some really interesting non LLM AI projects which I know and love but they were pre 2022 and google's price wasn't as much dictated by those projects as they are doing now y'know?

My conclusion is that A lot of people can't / couldn't invest into OpenAI / thus flowed their money into anything LLM / AI related in the markets... & the companies are loving this and mentioning AI as much as possible

Can we agree on this or not?

I can agree if you think that these companies are investing into infrastructure but that infrastucture is now mostly GPU's which are only really useful for LLM related tasks and becomes redundant for general purpose stuff like running servers for the most part.

Do we agree on this or not?

Also regarding infrastructure, The thing is, That most of them are just packing Nvidia Gpu's which is something that Nvidia also offers and others could do too but yeah, I can get that part but is it an "investment" is questionable...

Its an investment only if LLM's turn out to be profitable.

Firstly the cut throat competition means that literally everyone is competing in it so it cuts each other profits.

Secondly, there are some recent models which are kinda small and could run to a somewhat degree on modern hardware if need be which could satisfy some users needs without having to need that infrastucture

Then again, even if there are some people that might not have that and they search on things like chatgpt. They do it out of freebies that its not gonna cost them that much. And they can switch out if those AI providers do charge them first... with open source models while they themselves ride this end of AI hype.

If you believe that AGI is near, whatever that means, then literally everything I said gets out of the equation but I am assuming you aren't believing that.

Now sure there are gonna be returns but they aren't gonna be nearly as expected. In fact I think that most S&P companies are gonna be in a loss with all of these training of models / building infrastructure. Also, training of models is a recurring cost for the most part if they have to stay SOTA iirc with higher developer's cost working in AI/ML (100 million$ income is provided by the people investing into S&P dude)

So with all of these things, I believe that there is a legitimate concern that the investment isn't worth the return.

Then why are companies investing?

Because of fomo. When the AI hype started thanks to chatgpt. every private equity rushed for similars and that kinda leaked into S&P companies which are doing the same thing with AI hype mentioning it so much.

Do you agree?

If you can agree with all 3 of these statements to a reasonable degree, then I believe that we can agree that we are in an agreement and that it isn't much of an investment as its a way to somehow increase their stock prices by essentially mentioning the word AI and that's all that matters to them in the end, but its all on a proposition that someone is gonna buy the stocks thinking that its gonna go up and so on and so on.. when fundamentally the business model is kinda messed up when you think about it y'know? This is a bubble to me in my definition of it when people are investing into things without caring about things for the most part / logically I suppose...

If we have any disagreements, do let me know so that I can maybe lighten up on some other points as I love talking lol. I am loving it although I feel like I write realllly long sentences but hey, I am writing this to really explore why I believe the things the way I do and if you can convince me then sure, I can be wrong, I usually am.

Have a nice day and looking forward to your next comment!

▲

rapsey 7 hours ago | parent [-]

> My biggest issue which is the crust of this discussion might be that I believe that tech stock prices are roaring so high mostly because of the AI hype that they bring which raises their prices.

They are not that high at all, at least nowhere near bubble territory according at least to the financial analysts I follow. A better metric than simply p/e is looking at forward p/e because the current price reflects their forward guidance.

> My conclusion is that A lot of people can't / couldn't invest into OpenAI / thus flowed their money into anything LLM / AI related in the markets... & the companies are loving this and mentioning AI as much as possible

I guess, but that in itself does not mean it is a bubble.

> I can agree if you think that these companies are investing into infrastructure but that infrastucture is now mostly GPU's which are only really useful for LLM related tasks and becomes redundant for general purpose stuff like running servers for the most part.

Recommendation algorithms run on GPUs, which is a huge part of any social network (like meta and tiktok). Like I said there is more than LLMs and those need to run on GPUs as well. They also provide a service to rent out GPUs to other companies to run their own models and make a very good business of it.

> when fundamentally the business model is kinda messed up when you think about it y'know?

You are alone in that opinion. These are some of the most profitable companies in history, which is why they make such a huge part of the S&P. You are talking about a feedback loop of investing, which is normal in any investment bull cycle. It can turn into a bubble and we may be at the start of one, but being an AI skeptic investor just means not participating and having poor returns. The future is uncertain and it sounds to me like you are looking for reasons not to invest.

	▲	Imustaskforhelp 4 hours ago \| parent [-]
		Recommendation algorithms run on GPU's but they were running on GPU's previous to the AI hype and they are still running while running AI inference/training so those datacentres for AI are still gonna be vacated if demand drops down and so it isn't definitely something of an investment These companies have sort of saturated their markets and thus joined into LLM etc. to try to catch the new shiny thing. >You are alone in that opinion. These are some of the most profitable companies in history, which is why they make such a huge part of the S&P. You are talking about a feedback loop of investing, which is normal in any investment bull cycle. It can turn into a bubble and we may be at the start of one, but being an AI skeptic investor just means not participating and having poor returns. The future is uncertain and it sounds to me like you are looking for reasons not to invest. Please try to change the word Ai in this sentence to crypto to see how relevant it might be :> Also, this line kinda means "It may be a bubble but it pays right now" in the sense that you are basing your returns STILL on the fact of some predicted PE. I am just saying that people shouldn't consider S&P 500 "safe enough" then I suppose due to this AI hype if there is even a sheer possiblity of bubble formation. Higher profits generally mean higher risks and there is no free lunch. So S&P 500's higher profits does have a higher risk and people should know that risk before investing and my risk appetite doesn't support it and I am wondering how yours could. Superior returns aren't easy and if someone's saying them without giving you the underlying reason ie. realized productivity gains in an underlying trade (think a house builder built a house which was productive to the family and they are gonna pay for it) (compare it to how messy AI is, and how we haven't really still discussed on why there is so much hype in the market when the economy is doing kinda bad) > it sounds to me like you are looking for reasons not to invest. Yes, I naturally took the discussion from this side in investing in S&P markets and It's wild how you think so when I really agreed to you on a lot of things and your last line sort of sums it except when you look at the true gravitas of the situation, there is almost very little uncertainty about that (so no need for maybe) Sam Altman, CEO of OpenAI, has expressed concerns that the AI market may be experiencing a bubble, similar to the dot-com bubble of the late 1990s This is my opinion too. I was thinking of someone who wants to have a long time in the market as I think I said but time in the market beats the timing in teh market and so these "maybe" lines do frighten me. Do I want to mess around and find out if things are in a bubble with my money!? On a company which is massively enshittening itself in the names of AI (youtube auto dub comes to mind) This was a good faith discussion and I appreciate it but I don't agree on how it means not participating in bubble-ish maybe activities means you aren't getting returns, its like saying that I am not getting returns on crypto because I am not participating... because the whole thing is bubblish & those returns aren't magical... S&P should be considered a safe enough investment not something that is on the whims of a maybe, I suppose? I really like your last line I must admit, and it can take both an AI skeptic (AI skeptic in the sense that the tech is cool but its not gonna generate much profit given the investment) and pro AI person... > You are alone in this opinion .. I;d genuinely love to know if that's the case and I really wish to create an Ask HN, linking to this discussion as I don't think that my take is unreasonable?