AI really is going to be something 1.
What exactly? I have no idea. Theories range from:
But there’s another possibility I’d like to throw in the mix:
A useless tool we’re stuck with.
Right now there are many pitfalls to using generative AI in practice:
If we don’t find reliable solutions to these problems we just won’t be able to use it for critical work; i.e. the really useful stuff.
Consider those back office systems that took a small fortune to build. That money wasn’t spent for a slick flashy UI (and it doesn’t have one). It was spent so you could rely on everything running smoothly. Doing the right thing at the right time.
Could you run logistics, payment processing or legal oversight if your systems were built on injection attacks overriding your control? Or if every input had a random chance of just doing ‘the wrong thing’? And a new wrong thing every time at that?
But there are other uses, surely?
Sure. But remember when we got voice assistants? The potential for our Knight Rider sci-fi chatty-computer-side-kick dreams to come true?
9 years ago we got Alexa. Now we can to turn our TV on and off from slightly further away. For the low-low cost of potentially everything we say in private2. Amazingly it’s also costing the businesses running it a fortune; allegedlythey want to shut it down to save money.
We got stuck with a shadow of its potential. Halfway useful. Why? If I had to put my finger on it: investment.
Billions were spent on it. On the promise of it. Careers were made.
One thing we can concretely say about AI is that it’s getting a lot of investment.
Billions are being spent on it. On the promise of it. Careers are being made.
Nobody who rose up or is rising up on those billions is going to undermine their own ‘origin story’.
I don’t have a simple pill. But one thing we can do is, once we’re sure it won’t destroy us all, we can remember to make it useful, and only use it where it works.
Like any other tool.
If it’s not better than the status quo; keep cooking.
I feel it’s worth mentioning that I try not believe in tools and this extends to AI, LLMs et al. ↩︎
you can claim I’m wearing a tin-foil hat. But I’ve never seen proof to the contrary, and the potential is certainly there. Mic & network & power. ↩︎