Something came up in my mind as I was playing with GPT-4 today, watching it casually put together a smart contract line after line. Just as Microsoft was announcing the model’s upcoming release, we heard that it wasn’t really about wiping out jobs for humans. AI is here to help us with the routine, menial tasks, not to leave us pesky meatbags biting the dust. Those baby dragons in Game of Thrones were also cute at first, and look how that turned out.I am not saying it’s not cool that GPT-4 can build you a game within minutes or turn your hand-drawn sketch into a website. Neither am I saying that companies will fire their Web design teams en masse and stock up on pens and paper. The sky is not falling (yet), and the process we’re looking at will be more complex and more multi-sided than the doomsayers may think. But it’s not all sunshine and rainbows either.
ChatGPT: How much human is enough?
Consider this: Microsoft is gearing up to integrate ChatGPT, the mighty creation of OpenAI, across the board in its product suite. We’ve already seen Google and Facebook rush into the AI race as well, so we can expect AI-powered services to proliferate beyond belief, growing more powerful with every generation. Soon, generative AIs will be everywhere, generating business decks, writing code, creating CGI effects in movies and taking on hundreds of other tasks that humans used to do.
“Tasks” is a good word, an AI optimist would say, as it means other tasks, presumably more creative and exciting, would be up to us — humans — to do. Well, not necessarily. To see how that could play out, let’s take this amazing animated short that was made by shooting live footage and processing it with Stable Diffusion. The creators trained the model on stills from a specific anime show to mimic its style and added special effects for some extra oomph.
Now, let’s say we want to open a studio and shoot similar shows. We’d want to have our own style, though, because we want our work to stand out, so we hire an artist to draw us a bunch of stills that all bear the unique mark of human talent and creativity.
We train our model on it and there we go — we don’t need the artist anymore. We can produce our show cheaper and faster than traditional studios, and as long as we stick to this style, the human artist doesn’t have to be involved anymore.
So yes, for now, humans will continue to be involved; the questions are, how many humans, and how involved? A small startup will likely be happy to outsource most, if not all, of its content needs to AI, and it would probably make sense for larger companies too.
This removes a certain number of writer jobs, making the professionals in the space compete for room that grows tighter by the day. The same logic applies pretty much everywhere else, from coding to visual design, even accounting for the new “machine whisperer” jobs for people who are really good at telling AIs what exactly to do.
Old automation with new tools
We can celebrate the empowerment that comes from generative AIs — it’s now easier than ever to write a book, create visuals for a webpage or build a helpful script or tool. But this empowerment comes at the expense of the gradual dripping away of opportunities for people, even if it comes to such things as a one-off gig on TaskRabbit. As AIs grow more sophisticated, they will compete for jobs at higher levels, across more industries, to the point where human professionals will begin to fall off hard.
Furthermore, let’s not forget that besides the AIs coming for digital-focused roles, there are also the good old robots that came for manual labor. Again, their rise has been a more complex process than both critics and advocates would want you to think.
You can do it in a way that wipes out hundreds of jobs while creating a select few positions for high-skill engineers and code wizards, or you can actually do it in a way that saves jobs, as they did in Sweden. The scales don’t have to tip toward the former if you throw generative AI into the mix — but the temptation will be there.
ChatGPT and the future of AI workers
When you factor in that machines powered by AI are taking over manual labor, things take an even grimmer turn for the human job market. Take a drink-mixing robot arm, pepper it with ChatGPT, and you get a robotic bartender that can fix you a drink based on your mood or a whimsical description of what you’d want it to taste like. Or what about a gourmet pizza dreamt up by an AI and cooked by a robotic chef? Don’t forget to tip the delivery bot that drops it off, and spare a thought for the humans that could have otherwise done the job.
Pushing this idea forward, it’s not hard to envision a point where a construction company would have an AI (owned by one of the big tech names, who else?), build a detailed 3D model of a building and then let a swarm of construction drones bring it to life through the miracle of 3D printing. No humans involved, and very few humans paid, of course, but that’s only an afterthought.