Robotics for sure. Aka “embodied intelligent agents”, a very old term from robotics that sounds over hyped but fits perfectly now. Vast categories of labor will be automated. And not just by big expensive robots. Even cheap Roomba-like robots are already getting arms attached and filled with GPT level intelligence to do home tasks. We will have dirt cheap robots everywhere. Labor is going to be automated across almost every sector. Nothing is safe from automation. We’re going to see teams of humanoid robots doing unimaginable feats by 2040. Just like how people screamed AI would never beat a human at chess or make art, robots will crush every category. The demand for a worker that can work nearly 24/7, 365, for 1/5th the cost, who will never ask for a raise, or go on strike, or unionize, will be huge. Eventually they’ll make essentially zero mistakes, and be vastly more productive than a human ever could. People joke that we don’t need humanoid robots farming because we already have tractors optimized for the task. But they’ve got blinders on and can’t accept that even in highly automated factories, there are still humans needed to bridge the gap between more automated processes. Some of the least at risk physical labor are things that require true human connection. Like, a Brazilian Jujitsu teacher is going to be extremely far down on the replacement ladder. Even if it could be automated, people want to compete against other humans in that case. But I would bet by 2040 teams of humanoid robots will be building houses and replacing skilled craftsman at many places in the home building process. From doing the concrete pouring and foundation, to building the walls, electrical work, laying tiles in the bathroom, etc. Maybe not 100% robotic automation, but by 2040 most laborers at a home construction job site will be robots. And again, it won’t just all be big expensive humanoid robots. You have to understand that the explosion we’ve seen with GPT-like systems being integrated into everything and becoming vastly smarter is going to happen again with robots. Another development analogy to look at is quad copters. At first the flight software hadn’t been fully solved and they required highly engineered high end parts and motors to fly. Sometimes limited to flight in research labs. Now we’ve solved the problem so fully, you can literally buy a DIY drone kit for $50, and instead of using the included X shaped plastic body for the drone, you can get two sticks from the woods and duct tape them together, and glue the motors onto the ends, and it’ll still fly. The onboard computer and sensors will respond well enough to compensate for how garbage its body is. The same will be true for robots. We’re going to have cheap robots that cost as much as a vacuum, 3-5x smarter than ChatGPT is now, doing weird little tasks all over society.