AI and Work: What Happens When AI Can Do Your Job Better?

I appreciate Rachel Gilson's work. What I find amusing as I get older is that everything just recycles. The concern that people have around AI, and even the concerns specifically brought up here, were the same ones being brought up in the 1980's when secretaries and other jobs were being impacted by computers. I had secretaries for years. As the senior most executive in a company (besides the CEO), I haven't had a secretary for more than 10 years. I have managed thousands of people and more than a half billion dollar budget. The position in the 2000's would have definitely necessitated a secretary. But not today. We began to call them Administrative Assistants and then the position eventually died out. We talked about the value of a Secretary and validating that work. But in the end, the work they did was taken over by computers and automation. I don't need someone to manage my calendar (because it is automatically done). What that did was not create a higher value series of positions, like Chief of Staff and other roles that create even more value. AI will do jobs better. No doubt. But that doesn't mean you as an individual are no longer needed. Learn to leverage it, grow your skills and mature. There is even more opportunity for individuals posts AI than there was pre AI.

I now use AI daily as part of my job. I write low-level firmware/drivers for WiFi chips that go into phones, routers, specialized equipment, etc. The ability of AI to digest a 2-million line driver code base, understand the impact of various functions, and do a very good job with debugging is unparalleled in the years of my career. Just having AI read a debug log with hundreds of thousands of lines saves me huge amounts of time.

What I have found, at least in my field, is that AI is really good at collecting and collating data, seeing trends, finding bugs, and even writing some solutions to those bugs. However, at least at this point, you still have to take its findings with a grain (or more) of salt, as it still can easily misunderstand. It’s improving by leaps and bounds all the time, and as our company model learns from all the employees using it and thereby educating it, it’s getting uncannily good at some tasks. Just the difference between now and 6 months ago is huge.

What it does NOT do well is in tasks requiring judgment (like when should something be done vs. when it shouldn’t). I’m not saying it won’t get better at that too, but I suspect that exercise of wisdom (vs. knowledge) and judgment will likely elude machine brains for some time to come, if indeed that problem is ever solved. As foreshadowed in the 80’s movie War Games, we want to be sure that humans still are the ones making the judgments.

(As an aside, similar to what dgszweda said above about secretaries, what I have noticed is that tasks that used to be done by interns at my company, or very new engineers, are done much better with AI. I’m not sure what that’s going to mean for training eventual senior engineers, but I suspect it means my company will have to learn a new way to use new or lower-level engineers in a way that will eventually yield senior engineers, because the way we have used them in the past is now pretty much obsolete.)

Many tasks will go away due to new technology (does anyone here actually remember elevator operators, and if you do, do you really miss them?), but that only means humans will need to continue to learn skills that technology can’t replicate.