Today's AI is Alot Like the 2008 Financial Crisis

Power · July 8, 2026

The Artificial Intelligence systems that most people use today are what the technically-inclined people call Large Language Models (LLMs). They're the chatbots most of the pulblic use from companies like Anthropic and ChatGPT. To put it simply, they're predictive engines that require massive amounts of data to hone their predictions. By prediction, I don't mean these systems are like Nostradamus. They are built on mathematical modeling that predicts with stastical weighting what the next word in a prompt will be and what the correct answer to that prompt or question is. In other words, it's a system built on probabilities.

The designers of these systems would have you believe they are something more than math and circuits. With some such as ChatGPT's Sam Altman going so far as claiming these systems may even have consciousness. The only evidence he and others usually point to in this assertion is that the human engineers often don't fully understand the answers that the systems give or the way in which they make their conclusions. I suspect these claims are more about marketing than anything else. I mean, if people think you've captured indpendent intelligence in circuit boards then people will be clamoring to give you money to get access to that fantastical knowledge.

In reality, these systems are using trillions of calculations to assess petabytes of information to provide their answers. If the engineers that produced these systems actually could predict what will be produced and how the systems developed their output then I would question their product. In others words, surprising connections and answers from massive technological systems is to be expected. But the problem with these systems is that the answers they give are not always correct. And they're often not even correct about the most simplest of things. Numerous lawyers have been sanctioned by judges for using AI "hallucinated" cases in their arguments. This should be a huge red flag to us all since the people that are running with AI answers without checking are people with years of graduate level education training them in confirming sources. If this is happening to them, then you can be certain this is happening widely across all demographics. And especially where there aren't checks and balances like a judge or opposing counsel to set things straight.

Now that I've defined what AI is and established the AI problem generally, I guess it's time for me to deliver on the promise of my title and do some comparing with the 2008 financial meltdown. Alot of people use references to financial products like "sub-prime mortgages" or other euphamisms like "the housing bubble" to describe that event. Those terms, however, create the perception that the cause was substantially with individuals who bought homes they couldn't afford or were somehow to blame for wanting to own a home in the first place. In reality, the driver was the immorality (and illegality) of the top executives and CEOs in Ameria's financial industry. It's hard not to see deliberate action in all the steps leading to the collapse. There was enough foreknowledge about the damage that was being done that watchdog entities like accounting firms were incentivized to look the other way.

And what exactly were they being told to look the other way from? The primary source of the financial products, that's what. And nearly every major player had a role in the chain of action that created the orchestrated confusion. Banks are what made adjustable rate mortagages to individuals they knew couldn't afford the inevitable later adjustment upward. Investment firms were the ones who packaged those loans with "safe" loans. Those bundles were then sold to "institutional investors." The institutional investors making the investment decisions weren't the people who actually owned the money, however. The vast majority of the funds managed by "institutional investors" are actually public funds--retirement accounts from state pensions for instance--or from other pots like insurance premiums to be paid out should bad events insured against actually unfold.

And the result of all this? The government came in with taxpayer dollars and covered the costs while the people who took those improper actions made massive profits. There are many similarities between 2008 and the "Nifty Fifty" stock collapse of the 1970s and the hostile takeover era of financial raiding in the 1980s. But going into those other periods is beyond the focus of this article. What unites 2008 and the AI chatbot era of the 2020s is the mainstream push for people to accept the answer of the "expert" middleman without question. The middlemen in 2008--the bankers and investment firm CEOs--were experts with math on their side. Likewise, the AI platforms that give you answers to questions large and small are sophisticated pieces of technology--developed by experts with math on their side. But as 2008 showed, the math is only as good as what variables get counted in the equation.

With AI, the companies involved create rules and protocols that make the systems even more opaque than their structure already inherently make them. The data scientists and AI engineers impose safeguards and weight some information more important than others in terms of sourcing. In other words, there are biases within the systems that developers and company executives put in place. And because these systems are proprietary, you have no way to find out what those biases are. That's how a system can spit out a made up legal case with as much confidence as telling you the sky is blue. It's a human-made system and nothing more. And once you realize that simple truth, then you start treating it as it should be treated--with healthy skepticism and a good old-fashioned online search to double check its work.