Investors are suddenly getting concerned that AI isn't making any serious money
futurism.com
futurism.com
What does it matter that they have no idea where to sail it? They know they are bound for some port, and when they arrive it will be glorious.
A yacht salesman will always encourage you to "Sink your life savings in a boat !"
Some are just more seaworthy than others, size and conditions mean a lot but anything can go to the bottom where all is lost.
Wherever you end up, you can get there in style if you can afford it.
And some ships spread so much pollution that the waters become toxic wastelands... Just like we can witness with the introduction of AI to the Internet
Which could go without saying but I said it anyway ;)
Only on the sense that infinite money solves all problems.
There are two variables on that equation: how much money they have, and how much they need to spend to reach a useful product. The second one varies way more wildly than the first.
Where lies the land to which the ship would go?
Far, far ahead, is all her seamen know.
And where the land she travels from? Away,
Far, far behind, is all that they can say.
- Arthur Hugh Clough(The following is a general comment not directed at you OP)
The human trait of disconnected intellect always fascinates me. People will predict something very publicly, never shut up about it, and then loudly proclaim “I told you so” when it happens, and then I say “well, how did you act on this knowledge to benefit yourself” and I always get crickets.
It’s like people are too afraid to actually commit because if they are wrong it would imply their world view is wrong (and the intellectual gap between them and their couch is narrowing) but if they are right they can ignore that little inconvenient thing about their predictions having survivable bias and walk around with a smug sense of satisfaction. I have always thought it extremely odd how someone can proudly exclaim their predictions and then have nothing to show for them, humans confuse me a lot.
Only a small fraction of people have both of those. But supposing that you are one of the lucky few, how would you propose to go about profiting from a bubble like this?
Buying put options on NVidia & co.
Google's cash cows are pure AI. Is that close enough?
They do provide shovels too, but that amounts to only around 10% of their business. If they have to subsist on their shovel business the stock price will still plummet. You don't have to see complete bankruptcy in a company to win.
In my encounters, the people making money off the AI boom (that I feel is a bubble) are large entrenched corporations, and the AI start-ups that have flourished I've encountered had critical factors that would be impossible for me to replicate (Harvard/Princeston grads who have VC investment connections from their family).
The large corporations leveraging it haven't rocketshipped their stock price, and the start-ups that have really increased in valuation are all pre-IPO and aren't exactly inviting me to join in their series A funding rounds.
So all that's left to my observation to "act on this knowledge" would be to try to bafflegab some LLM snakeoil amongst the 1000s or 10ks of identical attempts, and honestly I don't see the LLM bubble being any more viable than any of the previous things (crypto, etc.)
So, is there something I'm missing on how to "capitalize on the knowledge" that the current AI trend is a bubble?
Given they have their AI company in their bio, I guess they thought better of it and deleted it (https://atomictessellator.com)
Given their reticence in actually addressing the comments, I would presume they're happy that the company they founded long before the LLM boom that leverages AI is getting additional attention and/or funding thanks to it, despite being completely unrelated to LLMs.
That's good fortune and catching the wake of ChatGPT thanks to the general public's ignorance on the tech. Nothing to warrant such a smug, condescending tone on his part.
I think your characterisation above of AI companies as either entrenched, nepotism or snake oil speaks a lot to how you perceive the world.
Yep! I could see the rise of AI so I put my personal money where my mouth was, just like I did with crypto before that, and fat/thin client computing before that.
And I do see the futures ahead of time and I do act on them with my own money, taking the risk upon myself.
Correct, I’m not going to share info on competitive zero-sum opportunities with strangers for no reason.
I’m not ashamed of what I’m saying, that’s why there’s public information on who I am and what I’m doing on my profile (unlike you), my last comment was very sloppy as it’s midnight here, I deleted it because I couldn’t be bothered going down the rabbit hole.
And yes, I find these threads (mostly people’s expectations in them) enormously funny.
Oh and a minor correction: we actually use LLMs a lot! They are hugely helpful for writing esoteric input files for quantum chemistry packages, and we also have a custom LLM trained to produce crystal structures (this LLM has already made a new discovery in electric battery tech that has been verified in a lab!), so “unrelated to LLMs” as you wrote is very wrong, we consume their output and produce entirely new crystal structure generating LLMs!
Just how far in the hole has that left you, seeing that you are left to resort to the free entertainment like the rest of us poors?
That might be inspiring for those who want to gain insight to success in unrelated technologies which they might have foresight into.
The easy money are for the entrenched.
Many already have, as they're the ones that sold to investors. And for any that chose to remain on the sidelines, not footing the risk can be pay off enough. Shorting however, isn't about predicting a bubble, but predicting if/when it bursts. And that's not an easy prediction to make as investors can choose to sit on unrealized losses indefinitely, if they're not over leveraged or needing to realize them.
Selling shovels during a goldrush is a great idea but not really feasible with AI unless you're NVidia.
Not saying it isn't a bubble, but people thinking it's one isn't evidence.
Same goes for the WWW. The dream of everyone keeping their own website full of useful information, all helpfully linked together, hasn't really materialized. As with before, a few people are doing that, but it never swept the world. The WWW for most people merely provides a way to load their games.
Nope, not at all. Relative to the people I know I'd say only about 10% have a personal computer for gaming, all the others use it for work or hobbies that do not involve gaming.
> Same goes for the WWW. The dream of everyone keeping their own website full of useful information, all helpfully linked together, hasn't really materialized. As with before, a few people are doing that, but it never swept the world. The WWW for most people merely provides a way to load their games.
My point has never cited anything about the WWW being skeptically viewed as a fad as the dream of everyone having their own website. That wasn't the view at the time, it was the view of companies not wanting to start a website because the WWW would be just a fad, it was people browsing the internet once, and saying they didn't want to waste time on something that would disappear in a few short years.
Narrow world perspective? Or narrow gaming definition? HN, for example, should be considered gaming. It even keeps track of your score, so clearly it is a game. When we look at the data of where people generally use their personal computers, gaming is it.
Yes, there are outliers. There always is.
No, it's not a narrow world perspective, I cruise around quite varied social groups, artists (music, video, performing arts, visual/fine arts), scientists, software engineers, blue collar workers, so on and so forth; most of them use a personal computer to actually perform a task: send emails, research in general, edit something (pictures, videos, text, etc.), plan projects (personal, hobby-related, or professionally), make art, the list goes on and on.
These are not outliers, these are normal people doing stuff with a personal computer.
I believe you are the one having a narrow view of how people use their computers, the only ones I know gaming on a personal computer are the software engineers. The other gamers in my sphere have a console for that.
It sure looks like a game. You write a string of text that otherwise serves no purpose to see if you can goad a reaction, and it even keeps score so you can keep tabs on your gameplay performance. If that's not a game, games must not exist.
Most of the rest of us post because we think we have something worth saying on its own merits, whether or not we get points for it.
If that were truly the case, why not simply talk to yourself in the shower or keep a private journal? HN would serve no purpose. The only thing it offers, that is not offered by a shower or private journal, is the gameplay.
No, for God's sake, it offers other people, sharing their knowledge or opinion. If that's not important to you then I have absolutely no idea why you are here.
Games cannot also provide knowledge or opinion...?
But you literally said you don't come for the reaction. Where is the knowledge or opinion coming from if there is no reaction?
> it offers other people
I see no evidence of other people. What people are you seeing?
Me neither ;)
I might as well be a bot myself for how little I get out of so much scanned material :)
>why not simply talk to yourself in the shower or keep a private journal?
You can't get much better advice than this. This is like a true gut response. These have been proven so effective for so long it can often be assumed that people are already doing it :)
And then some.
Well I like it. All kinds of people earn points from me for all kinds of reasons. I wouldn't want to downvote either without being motivated by something meaningful enough to comment on. So I'm one-sided and generous, enjoy it that way, and will ante up again.
But it's nothing like an actual game, there is no strategy for me and I'm not playing anything.
Literally, I spend dozens of hours reading articles and comments for every hour spent gaming, but so much of it falls into the category of amusement.
But people can and do "game" the system or play for points or only in ways geared toward "winning a popularity contest", that's them, I see that.
Looks like you're coming from different places which are all normal and have good "points".
Not that much different really, depends on where you draw the line.
At the extreme you could put video gaming, facebook, youtube, blogging, and anything slightly like that into the category of "amusements" unless it's entirely very serious with no amusement component at all.
It's another impossible line to accurately draw and probably a moving target at all times, but is it a big percentage? And are "serious" people getting their computing dollar's worth if so much is being diverted to amusement?
Or is it the other way around? Where those craving more of the amusement wondered why it took so long for the hardware & software to finally become widely capable of playing multiple video ads at once. While the real tech-types were doing nothing amusing at all, just trying to get their share of more computations out of computers whether it is entertaining or not.
What a way to shortchange those who thought they were due for greater amusement ;)
Some of them where the sky's the limit on how big a screen they will spend on, to amuse more boldly than the neighbors :/
I use Hacker News to educate myself and share my perspective of the world with others, and they do the same in reply to my comments. That's not a game, at all.
Your narrow view looks at it as a game, that's your opinion and not a fact. I don't use Hacker News as a game and I guess I'm in the majority on this here...
I'd guess it's the same kind of people blatantly ignoring everybody all around them not being able to use gen-AI and claiming it's a hit.
At least, they share a trait.
(But yeah, people have already found uses for gen-AI. Just not any of the uses that the people trying to claim a hit insist on.)
Some elephants are more difficult to see in some rooms than others.
It can depend on how good the bifocals are ;)
Or if the only thing that can block the view is piles of cash, it might take so much cash to be unattainable by almost anybody, but if it happened bifocals won't be able to help in that room, and the only thing that might be able to get any attention is a bigger elephant.
At least until the cash has dwindled :(
Most computer programs (almost all?) are not supposed to be worth money to begin with. It can require quite a reality-distortion effort to build positive sentiment on a case-by-case basis. As we have seen, after some people get that good at building sentiment, they can choose to do so without anything of value underlying their effort, whether integrity is intentionally being compromised or not.
The odds to be played here are one of the things that makes it more of a crap-shoot and defines an investment landscape where all the duds need to be offset by a single home run out of the park, but which one will be the heavy-hitter is never very well identified in advance of going viral. For that one outfit, it can be like printing money, for everyone else the default is a black hole.
AI of some kind (probably ramen kind) may be possible to leverage without putting a black hole onto steroids, but otherwise I could only imagine that if sustainable research was still required beforehand, the most suitable structure might just be a non-profit organization. Who would think of that unless there really weren't enough organic profits to be leveraged like a financially sound investment should have?
With some of the more challenging or nearly-unattainable projects, almost all approaches are not going to be profitable enough to provide a return for shareholder-style investment, most things like that are just too expensive to be justified except as an indulgence, which is where pride can be taken without anticipating a financial return. So money used to fund the effort is most accurately categorized as donations not investments and another reason that a non-profit structure, and resulting not-profitable outcome should be considered ideal.
Or maybe one particular approach is all wrong only because it is too expensive, overpriced, not-profitable or all of the above.
OTOH could be an approach is too expensive, overpriced, not-profitable or all of the above, because it is all wrong to being with.
But what difference does it make if the final outcome is the same?
I highly admire your success if 95% of the code you have written is worth something, especially money.
I could be wrong but I respect that number as an outstanding outlier, maybe I'm just underestimating what average professionals are capable of, but I would think you are still head & shoulders above most of them with stats like that.
95% of my code has ended up in the garbage bin, not even recycle, and I'm no professional so maybe I'm just grossly overestimating the number of pure software companies that failed without ever issuing a commercial product of any kind because their extreme programming effort was not able to be brought to a logical enough conclusion for monetization to take place.
I can be a dingbat sometimes but I'm completely in awe of the value of deeper experience and nearly-universal success like that :)
How well have you done when it comes to AI so far, and where do you see the money being made over both the short-term and long term where opportunities still exist?
If you would please not create accounts that do that, we'd be grateful.
> It can depend on how good the bifocals are ;)
It can also depend on how much your salary is tied to "AI being the next big thing." I think a lot of HN commenters are directly or indirectly working for companies that have hitched their horse to AI, and there is a lot of X-Files "I Want To Believe" here.
Actual article among others: https://www.washingtonpost.com/technology/2024/07/24/ai-bubb...
(Lots of discussion: https://news.ycombinator.com/item?id=41087719)
Maybe one starbucks late per year... just maybe.
Go do more carefully crafted demos to CEOs that don't try to use the product themselves and maybe you'll get some more sales.
AI should be a bit quicker, given its accessibility. Stubborn humans and their bureaucracy will hold that back a bit though. For instance my company is concerned about data privacy, needs to go through the legal departments approval then IT's approval, by the time it's in the actual hands of the user it's obsolete or limited in use case.
Definitely a big opportunity for investors, the long payback period is the issue.
We have at least one new major firm out of the initial boom, with bright prospects, and huge cash burn, but a TON of customers.
The tools are so general right now, there’s a lot to be written. It seems certain that we’ll get cost savings, automation, and almost certain that we’ll get new revenue streams out of this — whoever figures out how to get new revenue that’s not based on SAAS model delivery of cost savings is going to be very happy.
To keep the long view, the last time we had such broad individual mass market adoption of a technical tool regardless of corporate bans and rules was the iPhone, before that, probably Linux in IT departments in the 1990s. People WANT and NEED these LLMs for a lot of daily work. Some folks are going to figure out the business models and make some venture funds very happy.
Current level AI will be able to automate quite a bit, it just takes time to proliferate, there is a lot of inertia in organizations.
The reliability of models on longer, more complex chains of reasoning is the biggest issue with the current gen AI wave, in my view. Very interested to see how and if this changes over the next few years.
A huge amount of possibilities open up as this goes from 50% to 90% and higher.
When you can break down tasks into manageable units and then chain dozens of them together with a decent chance of the model successfully completing the entire chain, this could enable a big jump in capability.
AI isn't even like, 10% deterministic at this point. And that's kinda the point of the tools we're using.
But I also doubt if it will have that much effect. In general we have diminishing returns from these kind of technology, telegraph's effect on the world was larger than the computer, computers were more valuable than WWW and search engines have smaller effect than WWW. AI may be big, but could it be bigger than search or e-commerce? I doubt it.
https://en.wikipedia.org/wiki/Productivity_paradox
If computers really had little impact on productivity then things like the Cloudstrike outage would be barely noticed, instead of causing instant mass disruption worldwide due to sudden productivity loss. It would also imply that everyone, everywhere, in every industry, is an idiot who adopted computers despite not getting much return. That isn't plausible, so it's more likely that there's a problem with data.
In particular, note that productivity growth of 1% is compounding. Each year you need bigger and bigger upgrades to sustain that 1% growth.
That is a massive amount of content and products you consume today that just aren't accurately valued by GDP, and thus not captured by productivity measurements. The companies providing those services makes money in other ways, sure, but that value isn't even close to the amount of money it would have costed to buy that much products for consumers in the old economy.
Just take the amount of porn consumed by the average teenager today, that would have costed a fortune 30 years ago while today he got it for free.
Maybe I'm just overly optimistic, but on a wider society scale I see a lot worse things they could be pouring money in to (eg Ads). Even if they never make money on it, at least we end up with a lot of open knowledge through all the papers that have been published and even some open-source models that can do some neat things.
And even if the academic side is a small amount of the funding, it's at least something. Some of the rest is being spent training models I can download and run forever more.
I guess mainly I'd rather focus on the interesting technical and scientific side of AI rather than worry about the investors.
All that needed to happen for dotcom to workout though was high speed internet that business had come to people's homes.
For this bubble to workout we need to come up with artificial beings on the level of humans from limited use case LLMs.
It is completely delusional.
Robots?
We've got loads of those already. Ones with super-human speed, super-human dexterity, super-human senses. Even got some cute tech demos where they get controlled by the OpenAI API.
I'm not sure why you think that's delusional.
But I'm also not sure why you think that's necessary, given that not only "all things which can be expressed in written form" can be used by an LLM, but also that the same architecture is fine with anything that can be tokenised which includes visual and audio tasks — i.e. most office jobs.
No, what's necessary for this to not be a bubble is for someone to figure out how to turn what exists into an actual business model.
"Business" is much less flashy, less cool, less exciting than tech demos filled with androids doing housework, parkour, or acting as a tour guide.