Analysis says that AI added zero to the economy in 2025. Is this alchemy (turning silicon into gold) or the quest for unobtanium?

Photo below - when completed, this $2 billion data center near Columbus Ohio will be the world's largest. Google is contemplating a total of 3 data centers at this site alone.

Suppose you invested $2 trillion, and made zero profit. Okay – lots of startups in their early stages do that, right? Facebook, Tesla, TikTok, Theranos, IRL. But usually, they have SOME net cash in-flow from sales, and their labor force impact is detectable. Well, maybe we should scratch IRL off that list. It imploded (2023) as soon as sleuths revealed that 95% of the site’s 20 million users were “bots”. A Billon dollars in IRL market cap was erased overnight. Investors never got a dime back.

But the failure of startups like “IRL” has nothing to do with why AI may have minimal economic benefit last year. It’s simply that economists haven't been able detect any revenue, expense savings from corporate clients, or net change in corporate staffing. (See futurism link below).

This isn’t the first time Wall Street skeptics have speculated that AI is on track to overpromise and under-deliver. But not maliciously so. More on the order of a mass hallucination. The kind that of hallucination which AI itself uses to keep faking photos of people with 6 fingers, citing research papers that never existed, and losing hundreds of thousands of dollars during their first hour of crypto trading, despite being trained on a robust data set.

Hallucinations are a form of mental illness resulting in self-deception. A better explanation of AI’s failures has recently suggested as “confabulation”. Merriam Webster defines confabulation as “filling in memory gaps by fabrication.” Machines are incapable of LSD style trips or hearing the voice of god, so far as we know. So confabulation – the desperate but deliberate creation of fictional facts to support a world view - is AI’s problem. Is that possibly the case with the giant corporations throwing trillions at AI too? They all believe that there will be just a couple giant AI winners/survivors. Becoming a winner could depend on building most, largest, fastest and KwH hungry data centers. The winners are counting on competitors to crash and burn because they didn’t build giant enough giant server farms fast enough. A cold war style arms race, but with chips instead of uranium.

If you want insight as to how corporate America thinks the AI future will play out, consider my daily experience. Whenever I start a new Microsoft Word document, the first thing it demands is “tell Copilot what you’d like to draft”. This happens by default- before I tap the first key. I can turn Copilot off for that specific document, but with the next document it happens yet again. If there’s an app-wide way to shut down Copilot completely, I haven’t found it yet. Microsoft may believe its survival depends on giving Copilot away at no extra cost for now, and then implementing fees when users become dependent. And who knows? They might be right.

I’m just sayin’ . . .




https://www.msn.com/en-us/money/markets/you-ll-snort-laugh-when-you-learn-how-much-ai-actually-added-to-the-us-economy-last-year/ar-AA1ZeTjK?ocid=BingNewsSerp

https://gizmolead.com/entrepreneurship/irl-closes-bot-scandal/