The first time I really considered the impact of AI on the workforce was at a meet up that we used to have called Silicon Hollar. This was started by a handful of guys, one of which had written a paper on how computers would affect Watauga County in particular. He estimated that due to the automation of machines, we could lose up to 40% of our jobs. I was thinking about things like self-driving trucks, robots restocking shelves at grocery stores, machines flipping burgers at McDonalds, you know, things that we could be doing but just aren’t. I’m always amazed when I see a video of a robot making fried rice in China, thinking that’s an easy way to replace a person, but tech jobs would always be safe, right?

I do believe that these jobs can be replaced, and probably will be at some point, but there are real hurdles to getting robots into our everyday lives. They’re complex machines that require maintenance. They break down. They’re expensive. They are hard to design and can’t really be used for anything except what they’re made for. We also have government regulations that keep them from potentially hurting people that you have to get through. There are large companies that are willing to take these risks and are slowly deploying robots, but there is a high cost of entry.

What doesn’t have a high cost of entry are the new generative AI products like ChatGPT, Copilot, Claude, and the like. Sure, it costs a ton of money to get a data center built and fitted out, but if I want to try out AI in my job, it’s $30 a month for a license to use all I want. Once I have seen what it can do, I may or may not choose to invest in a development team to make it more applicable to my needs. But the cost of entry there is still a lot lower than hiring a team of engineers to build a robot for me. Plus, we currently have zero government regulation on AI, so you can move as fast as you can to get it spun up and deployed.

This got me thinking, if generative AI gets really good, which it will in the next 1-2 years, what could it replace? Here are a few real-world situations that my friend and I came up with:

1. Currently App State employs two security analysts to comb through risky events to see if they are real, write up reports on them, and then proceed to turn off accounts or reset passwords when needed. We are already seeing AI security products that will do most of that job, and it won’t be long until they can do almost all of it. Those two could easily become one, if not zero in the next two years.
2. My friend manages a team of eight administrators that run a whole slew of servers. A lot of their tasks are around patching software, deploying operating system updates, making things work when they break, etc. In the next two years, that team of eight could easily become five, or maybe even less depending on how good AI gets at fixing issues.
3. On my team, I could see a world in which Copilot is even more deeply integrated into Windows and other connected systems and is able to solve issues for our clients directly. Maybe you want to update QuickBooks, just ask Copilot and it does it for you. Have a computer that isn’t fast enough, just ask Copilot to fix it. Can’t print to your printer? Copilot can get that for you. If Copilot doesn’t have the answer, could it ask another AI for assistance? I don’t see why not.
4. Outside of tech, I am going through Watauga Leadership Development this year, and we recently had a session on coaching. As I drove back to the office, I thought to myself, “AI could be a great coach.” Sure enough, there are already custom AI programs for this, and they’ll only get better, potentially putting a lot of human coaches out of work.

There are a lot of places that AI doesn’t fit yet. AI can’t plug in a computer or run cables. AI can’t put together a desk or replace a busted pipe. AI can’t build a house. But I think that we’re going to see AI affect a lot more people than anyone expects, and it’s coming very soon.