The AI wave is so reminiscent of the early days of the internet. Right now we're in about 1999, a time of tremendous hype. Business that weren't even in tech were wondering how much they need to know about the HTTP spec. People from the marketing team were being pulled off their print ad campaigns, given a copy of Dreamweaver, and told to make the company website. We hadn't even figured out the roles.

It 100% turned out to be a bubble and yet, if anything, the internet was under-hyped. The problem in 1999 was that no one really knew how it was going to play out. Which investments would be shrewd in retrospect, and which ones would be a money pit?

When an innovation hits, it takes time to figure out whether you're selling buggy whips, or employing drivers who can drive any vehicle.

Plenty of companies sunk way too much money into custom websites back in 99, but would we say they were wrong to do it? They may have overspent at a time when a website couldn't justify the ROI within 12 months, but how could they know? A few short years later, a website was virtually required for every business.

So are companies really seeing "zero return" on their AI spend, or are they paying for valuable lessons about how AI applies to their businesses? There may be zero ROI today, but all you need to do is look at the behavior of normal people to see that AI is not going anywhere. Smart companies are experimenting.

The internet is about sharing actual information. LLMs can't share real information, just a flimsy derivative. It's in their DNA. How so many supposedly smart people who understand machine learning and its fundamental entropy failed to acknowledge this reality from day zero, is beyond me.