The State of AI
Over the holidays I spoke with several of my relatives who knew that I did computer things. The conversations always seemed to somehow transition into
andWhat do you think about that new AI thing?
Thinking about my own interpretation of where AI is currently at, I wanted to give my general feeling of the climate, and the probably-completely-wrong forecast for the short term.I read this article published by such-and-such that said everyone was going to lose their jobs.
What is AI used for?
Related to what I do, software engineering, people use it to make their day-to-day easier. They set up agents, which are like little coding helpers, to do different tasks for them. Agents are capable of doing as much as you give them access to, and as much as their knowledge allows.
The agents are backed by LLMs (large language models) which are big computer brains trained on internet drivel. They ingest all of the academic whitepapers and conspiracy theory nonsense, and are able to hold a conversation about those things when prompted.
Will it put you out of a job?
Maybe, but I hope not. My idea is that we should embrace this and evolve with it, rather than nay-saying and pushing back. It's one of those inevitable technologies at this point, and the most prudent thing to do is to learn as much about it as you can to not get left behind, and to more easily divine how it's going to change the engineering landscape.
I'm technically a manager now, but I like to still pretend I'm an engineer so that I can empathize with my direct reports, and understand the technical challenges they face - again in the vein of better predicting where we are headed.
What about all the recent layoffs?
LLMs are understandably impressive. They're impressive to C-suite execs, and the engineers using them. A lot of this hype has, in some cases, led to overestimations in future impact for companies. They pre-empt layoffs, thinking an LLM can solve all their issues