Betteridge's law of headlines is an adage that states: "Any headline that ends in a question mark can be answered by the word no."
It is based on the assumption that if the publishers were confident that the answer was yes, they would have presented it as an assertion; by presenting it as a question, they are not accountable for whether it is correct or not.
> [hallucinations therefore] AI can never be accurate or trusted enough to majorly disrupt the labour market through automation or augmentation, and this pool of revenue is never coming their way.
Humans drive really shit but we still let millions yolo it daily. You don’t need perfection to roll stuff out in real world, just good enough
The industry will certainly have to pull a rabbit out the hat here but I don’t think it’s a foregone conclusion that it won’t. Not so long ago the notion of revenue from language models wasn’t a thing, now it’s billions yet people can’t picture that number growing rapidly in future?
Seems like a well written article and after I read it, I can't help but think there is definitely a bubble.
The question is... as someone who's invested in the stock market, what do you do? If the bubble bursts, I don't want to be in stocks. Do I want to be in bonds? Other than shorting the market which I'm not keen on doing, what's the best way to profit from the bubble bursting?
My approach is to invest for the long term, diversify a year's worth of living expenses into assets inversely correlated to tech, hold a diversified portfolio across different tech-oriented future scenarios and attempt to take a systematic, unemotional approach in domains that I understand and where I have expert knowledge. Tech and NVIDIA in particular have seen very significant draw-downs in the past and yet have been (in my experience) good investments since I started in 2008. Specifically, NVIDIA has seen extended draw-downs in the 60% to 90% range multiple times in the past. You just have to be (financially and emotionally) prepared for the ride and don't imagine you can time them. Anecdotally, Berkshire plays the safe haven / inversely correlated role for me though I've not proved it. There are always segments in tech doing well so I research beyond the Mag 7 and infrastructure. See bio for my research. [My opinion and not investment advice].
> The question is... as someone who's invested in the stock market, what do you do?
Nothing, unfortunately.
The problem is that the bubble is a stretch of time while the burst is an instant. So, you lose all the benefits of participating in the runup because you exited at the wrong time while you catch all the downside of the burst because it took out all the markets simultaneously.
While I think there is an AI bubble I also think that when every big AI provider is claiming on earnings calls that they can't sell enough tokens to meet demand that the music will keep going at least a little longer. Combine that with positive unit economics for inference and there is at least a plausible business case that AI labs can make money soon. Anthropic claims as soon as the end of 2026 we'll see about that.
There is also a chance that a lot of this capex is written off, and the money becomes "sunk". Bad for the current players, but given inference costs as you mention are profitable, after the writeoffs and the market correction the industry continues on variable inference revenue.
The catch is you probably only want to be invested after any writeoffs/corrections if that is your hypothesis. i.e. the future may be AI, but it isn't a straight line, nor is it guaranteed that the current players will be the future AI company of choice. You can be right about the end state and still lose your shirt in between with markets.
> Essentially, the amount of human labour needed to identify and correct these AI hallucinations is greater than the human labour saved by deploying the AI. As such, AI isn’t even a widely viable option for augmentation, let alone automation.
I have been saying variations of this across all social media platforms for the last six months, and every time I get savaged by tech bros. The pro-AI ideology absolutely insane.
Betteridge's law of headlines is an adage that states: "Any headline that ends in a question mark can be answered by the word no."
It is based on the assumption that if the publishers were confident that the answer was yes, they would have presented it as an assertion; by presenting it as a question, they are not accountable for whether it is correct or not.
Except, and you would know this if you skimmed the article, this headline query is being used sardonically.
> [hallucinations therefore] AI can never be accurate or trusted enough to majorly disrupt the labour market through automation or augmentation, and this pool of revenue is never coming their way.
Humans drive really shit but we still let millions yolo it daily. You don’t need perfection to roll stuff out in real world, just good enough
The industry will certainly have to pull a rabbit out the hat here but I don’t think it’s a foregone conclusion that it won’t. Not so long ago the notion of revenue from language models wasn’t a thing, now it’s billions yet people can’t picture that number growing rapidly in future?
https://www.nytimes.com/2025/11/30/technology/david-sacks-wh...
Seems like a well written article and after I read it, I can't help but think there is definitely a bubble.
The question is... as someone who's invested in the stock market, what do you do? If the bubble bursts, I don't want to be in stocks. Do I want to be in bonds? Other than shorting the market which I'm not keen on doing, what's the best way to profit from the bubble bursting?
My approach is to invest for the long term, diversify a year's worth of living expenses into assets inversely correlated to tech, hold a diversified portfolio across different tech-oriented future scenarios and attempt to take a systematic, unemotional approach in domains that I understand and where I have expert knowledge. Tech and NVIDIA in particular have seen very significant draw-downs in the past and yet have been (in my experience) good investments since I started in 2008. Specifically, NVIDIA has seen extended draw-downs in the 60% to 90% range multiple times in the past. You just have to be (financially and emotionally) prepared for the ride and don't imagine you can time them. Anecdotally, Berkshire plays the safe haven / inversely correlated role for me though I've not proved it. There are always segments in tech doing well so I research beyond the Mag 7 and infrastructure. See bio for my research. [My opinion and not investment advice].
Treasuries.
Take a steady guaranteed 4% and sleep tight.
"Safe" bonds are less than treasuries, and a big funder of the AI bubble is bonds , so you will be they one holding the bag when they bust.
> The question is... as someone who's invested in the stock market, what do you do?
Nothing, unfortunately.
The problem is that the bubble is a stretch of time while the burst is an instant. So, you lose all the benefits of participating in the runup because you exited at the wrong time while you catch all the downside of the burst because it took out all the markets simultaneously.
diversify
While I think there is an AI bubble I also think that when every big AI provider is claiming on earnings calls that they can't sell enough tokens to meet demand that the music will keep going at least a little longer. Combine that with positive unit economics for inference and there is at least a plausible business case that AI labs can make money soon. Anthropic claims as soon as the end of 2026 we'll see about that.
There is also a chance that a lot of this capex is written off, and the money becomes "sunk". Bad for the current players, but given inference costs as you mention are profitable, after the writeoffs and the market correction the industry continues on variable inference revenue.
The catch is you probably only want to be invested after any writeoffs/corrections if that is your hypothesis. i.e. the future may be AI, but it isn't a straight line, nor is it guaranteed that the current players will be the future AI company of choice. You can be right about the end state and still lose your shirt in between with markets.
I assume much of the insatiable demand for tokens is from free users.
> Essentially, the amount of human labour needed to identify and correct these AI hallucinations is greater than the human labour saved by deploying the AI. As such, AI isn’t even a widely viable option for augmentation, let alone automation.
I have been saying variations of this across all social media platforms for the last six months, and every time I get savaged by tech bros. The pro-AI ideology absolutely insane.