A more pragmatic look at AI

It seems like you can't go on LinkedIn without seeing the majority of you feed talking about AI. There seems to be strictly two camps. AI will either: 1. Progress society to a utopia. 2. Destroy us all.

Like with most things, the truth is probably in the middle.

I'll be the first to admit I lean a little more in the naysayer territory. There are a lot of downsides to this emergent technology. There is a lot of hype (some of it deserved, a good portion not) but the hype is mostly driven by people with stock positions in its impending success. Most tellingly, this tech is ignoring one of the most basic principles of programming. Performance matters. The hype on these sophisticated autocomplete machines is starting to work because demand is driving the demand for yet more of these inefficient data centers.

Venture capital is paying the bills and wants more data centers, so company owners (unsurprisingly) build more data centers. It's not their money so why would they not build more. That's not to say the tech isn't useful (it is), but so was the internet. The current buy in for this tech isn't unlike the early 2000s. Every company wanted a website (whether they needed one or not). Once they went online their stock prices went up. The other day a shoe company rebranded as an AI one and their stock shot up 582%. How they differ from competition aside from just bolting a generic AI service onto things is yet to be seen. This is not sustainable growth.

When the hype falters and venture bills start coming due, the product will have to adapt to be more efficient, which will free up compute power to actually make strides into legitimate AGI that will change the market.

For better or worse that is going to happen. Industrialization pulled children away from the assembly lines. Punch cards let computers compute at the command of man. Code compilers came around and removed the need for punch cards. Card punchers still had jobs, just different. The mouse, excel, and drag and drop web site editors all had their part to play, but ultimately the people whose jobs were replaced specialized deeper into the new tech while previous technological advances were made accessible to the common man. With every abstraction layer comes a mass movement in the market and the ones who adapt are the ones at an advantage.

We can go all in on using AI (without actually learning how it works) or doom and gloom and want nothing to do with it; either way, the people pushing the envelope focusing on the performance and training the models how to convey lack of certainty in an answer before they are challenged are the ones who will get ahead.