<
https://freedium.cfd/https://wlockett.medium.com/the-ai-bubble-is-about-to-burst-but-the-next-bubble-is-already-growing-383c0c0c7ede>
"Speculation rules the world. It didn't used to. But from the 1980s through to
2008, something changed. Investors realised that they could get far more return
from hype than from any kind of legitimate business. This is the information
age, after all, and information is easy to manipulate and commodify. This led
to the dot-com bubble, the 2008 credit crunch, the 2016–2017 cryptocurrency
bubble, the late 2020–2021 cryptocurrency bubble, and the 2022 NFT bubble, with
the latest fad being the AI bubble. In fact, nearly half of the world's private
investment is being funnelled into AI, and AI speculation is the main driving
force behind the S&P 500's recent growth. But, just as the others did before
their catastrophic failure, the AI bubble is showing signs of imminent
bursting. However, the finance and tech bros have learnt their lesson and are
developing the next bandwagon to ride off into the sunset with all our money,
ready for when they inevitably need to jump ship. It's just a shame that it's
even more of a dead end than AI.
So, it's essentially common knowledge that the AI bubble is ripe for bursting.
Things like the efficient compute frontier and the Floridi conjecture mean that
the AI models we have now are about as good as they will ever be. Even if
OpenAI spent trillions of dollars increasing the size of their models tenfold,
they would only be slightly better. The recent release of ChatGPT-5 is a
perfect example of this. It had significantly more data, training, and cash
shoved into it than its little brother ChatGPT-4, yet it is only marginally
better than ChatGPT-4.
This is a huge problem! Because, as they currently stand, generative AI models
aren't actually that useful or even remotely profitable.
An MIT report found that 95% of AI pilots didn't increase a company's profit or
productivity. For the 5% in which it did, the AI was relegated to back-room,
highly constrained admin jobs, and even then, there were only marginal
improvements. A METR report found that AI coding tools actually slow developers
down. The inaccuracy of these models means they repeatedly make very bizarre
coding bugs that are highly arduous to find and correct. Logically, it is
quicker and cheaper to get a developer to code it themselves. Research has even
found that for 77% of workers, AI has increased their workload and not their
productivity. As it stands, generative AI is far too error-prone to deliver
meaningful increases in productivity or profitability in the vast majority of
use cases.
In other words, for AI models to actually deliver on the speculation driving
their massive investment, they need to become far, far better, which involves
spending exponentially more money."
Cheers,
*** Xanni ***
--
mailto:xanni@xanadu.net Andrew Pam
http://xanadu.com.au/ Chief Scientist, Xanadu
https://glasswings.com.au/ Partner, Glass Wings
https://sericyb.com.au/ Manager, Serious Cybernetics
