The Looming AI Crash, False Velocity and Reality. Vol 1 - Intro

This series of articles will talk about the reality of GenAI (Generative AI). The big GenAI companies are OpenAI and Anthropic, but GenAI is most any company using a chatbot that runs on LLMs (Large Language Models) We’ve been sold a bag of defective goods based on wishes and desperation by people with questionable morals who have a thirst for power. The ideas presented are not mine. They come from academics, analysts and experts. I’ve coined the term False Velocity to describe the game GenAI companies are playing.

False Velocity: The artificially amplified narrative that GenAI development is in an urgent, winner take all race, using fear to create pressure to move faster than is sustainable, primarily to benefit commercial investors and companies at the expense of safety, coordination, and the general public.

GenAI companies will likely crash in the near future. They’ll take the economy with them. There was a trillion dollar + correction in the last week! LLM’s have hit a wall and are no longer scaling exponentially, a requirement for them to achieve profitability. As LLM’s get larger, they don’t become more reliable. A billion dollars in current investment with no hope of removing hallucinations! A majority of researchers say that the current model and methods won’t lead to AGI (Artificial General Intelligence), a requirement for GenAI to do the work of humans and transform the world. Most of what you hear from GenAI companies and their industry is marketing and hype, look to researchers and analysts for the reality. Researchers have been predicting this scaling problem for decades.

”The majority of respondents (76%) assert that “scaling up current AI approaches” to yield AGI is “unlikely” or “very unlikely” to succeed, suggesting doubts about whether current machine learning paradigms are sufficient for achieving general intelligence.” 2024 report from Association for the Advancement of Artificial Intelligence.

80% of companies that have heavily implemented GenAI report little to no impact in productivity or hiring. GenAI companies and their investors are placing a six trillion dollar bet on a dead end technology that is currently making incremental improvements at best. OpenAI has failed to meet it’s revenue and user acquisition goals several times in the last few months and has stated they won’t be profitable until 2030, when investment in GenAI will have reached six trillion dollars. There is no evidence that they will achieve their goals. They are operating on the faith of a UFO conspiracist. These companies and their investors are riding on an out of control train, barreling off into the future. That train needs to keep increasing its speed, faster and faster, month after month to arrive on time in 2030 and it’s already near terminal velocity! These companies are pursuing a single strategy in the hope of achieving their goals, LLM’s. They’ve placed a bet on a winner that can’t win with no time or ability to take a different path. They can’t iterate, they’ve gone all in on a failed strategy.

OpenAI hype has fallen into a predicable patterns that we can see in Anthropic’s latest release, Claude Fable 5: Scare, Hype & Release. This has been the game book since GPT-2 in 2019, which was “Too dangerous to release.” Complete bullshit. Get peoples attention by causing fear, make outlandish claims about world transformation and mass unemployment, to drive up user adoption and sales. Not to mention valuations. They have created false velocity, there is simply no race to win given the limitations of LLMs. These systems aren’t smart enough to be dangerous, as they are not intelligent at all. They are word prediction calculators. Would you use a normal calculator that was wrong 10% of the time? If LLM’s were headed to AGI, that would be a reason for a race. They are not!

Unlike traditional datacenters, GenAI datacenters can’t be repurposed, when GenAI crashes there will be over a billion in capital expenditures on hardware that will be scrap. Launching of new datacenters has stalled, not a single major datacenter the broke ground in 2023 has gone live.

Building more and more datacenters with little regulation and no environmental impact studies, no community input or regard for the communities they are built it, benefits no one. Letting these companies grow without regulation is foolhardy at best, and there is no reason to do so, other than profit for investors.

Very few jobs are being lost to GenAI. GenAI can only do 2.5% of the work of any knowledge worker, and the idea that they can do the work of a blue collar worker or be commercial driver is a far off dream. Forget about self driving trucks. This is GenAI companies using fear to control the narrative, to get people to accept that GenAI is inevitable and will transform the world. There is zero evidence that this is happening. The claims have no basis in reality, they were made to create fear and grab attention. Every time a GenAI company makes a claim about danger it’s been a lie. GenAI has caused harm. Anthropic shared its codebase, there have been isolated incidents of ChatBots encouraging or leading to psychosis and suicide, and no one warned of these real dangers. Imagine a therapist’s session that resulted in talking the patient into suicide? They’d be held responsible, there’d be an investigation. But GenAI has no agency, it can’t take responsibility, but its creators can and should be!

This is the first in a series of articles where I’ll present research and discussion from experts in AI, finance, technology and the like. I’ve made some bold claims, and the evidence will be clear in the coming weeks. My sources have made bold claims, and I’ll provide the citations to back them up in the following articles.