We're only a year-ish into the explosion of generative AI. From popular usage to tech companies, it seems that everyone who's anyone is using it or has an opinion about how it's the next Industrial Revolution in the making or is going to put a million people out of work. I honestly don't know what to make of it. But I think that lots of things are going to change and that we can't really predict what those'll be. AI has certainly got me thinking about whether software engineering is even valuable anymore, or if ANY white collar desk job is. Questions that no one has the answers to — not even OpenAI.
What I have noticed is that AI lacks a certain degree of nuance. After all, it is a probabilistic model that outputs the most likely string of things based on an input. It therefore puts out answers that make sense, on average, based on all the data that exists. Coming up with a successful novel or business, then, is something out of reach of generative AI; things that succeed in certain domains depend on originality and subverting the average, not following what makes most sense. I wonder if we'll ever run out of novel human material to train AIs on when everyone's using it? A convergence onto a single answer means that all the variations and details, which might actually constitute the whole truth, are lost.
In a similar vein, human tasks that can be done by a robot will be done by a robot. The sort of repetitive knowledge work that an AI can do as well as a human, or better, is fucked. With the low-hanging fruit of work plucked, it seems that true expertise in a domain will become critical to making sense of anything that the machine outputs. Sure, the AI might be able to write an entire application at one point but maintaining the code over years of team changes, migrating the service from the cloud to home servers based on costs and priorities, and similar subjective decisions will ultimately come down to humans with opinions and expertise in the domain they're working in.
AI is exciting, scary, and disorienting. I feel that people should be worried about it more and thinking of it less at the same time. There's going to be many new opportunities that emerge from it that'll require NOT listening to the "experts." I'm hoping for an era similar to the early days of the web, where anyone with a laptop could make magic technology with limited entrance costs and share it with the world. I doubt that anything will ever beat the miracle of the open web, but I sure hope that AI comes close.
What I have noticed is that AI lacks a certain degree of nuance. After all, it is a probabilistic model that outputs the most likely string of things based on an input. It therefore puts out answers that make sense, on average, based on all the data that exists. Coming up with a successful novel or business, then, is something out of reach of generative AI; things that succeed in certain domains depend on originality and subverting the average, not following what makes most sense. I wonder if we'll ever run out of novel human material to train AIs on when everyone's using it? A convergence onto a single answer means that all the variations and details, which might actually constitute the whole truth, are lost.
In a similar vein, human tasks that can be done by a robot will be done by a robot. The sort of repetitive knowledge work that an AI can do as well as a human, or better, is fucked. With the low-hanging fruit of work plucked, it seems that true expertise in a domain will become critical to making sense of anything that the machine outputs. Sure, the AI might be able to write an entire application at one point but maintaining the code over years of team changes, migrating the service from the cloud to home servers based on costs and priorities, and similar subjective decisions will ultimately come down to humans with opinions and expertise in the domain they're working in.
AI is exciting, scary, and disorienting. I feel that people should be worried about it more and thinking of it less at the same time. There's going to be many new opportunities that emerge from it that'll require NOT listening to the "experts." I'm hoping for an era similar to the early days of the web, where anyone with a laptop could make magic technology with limited entrance costs and share it with the world. I doubt that anything will ever beat the miracle of the open web, but I sure hope that AI comes close.