There’s been a lot of digital ink spilled over generative AI and its impacts in our current moment. It is (at least for now) the biggest thing being pushed into all areas of our lives. AI-generated ads fill up our social media. Thousands of videos are being generated every day, sometimes for amusing jokes or skits, other times as weapons of disinformation and propaganda. Companies in all industries are looking to GenAI to cut costs (which mostly means laying off humans). It simultaneously feels like an inevitability and an artificially induced push meant to legitimize and justify the technology.

As a student and a software developer, I have seen GenAI evolve in over the years with great interest. For the past two years, students in every class I have taken have been using it, either permissibly or not to, do complete assignments, cheat on exams, and pretty much anything else you could imagine. I have also been privileged enough to learn about the internals of large-language models in the computer science electives offered by my university. Policies at UTK have changed significantly, with many of the initial restrictions on AI usage relaxing as time has progressed. At work, the push towards GenAI has been more in the industry as a whole rather than in the office. It’s becoming more and more apparent that software engineers who push it off are delaying the seemingly inevitable.

My thoughts on GenAI have been changing a lot recently, which is why I wanted to write them down. For the longest time, I’ve tried to push it off and regarded it (and the people who use it) with a lot of scorn and cynicism. And much of that still holds. I worry about the carbon footprint of these models, how much data is being pushed into a seeming endless number of black holes that suck away all our money, attention, and GPUs. But much my initial feelings were based out of fear. I felt (and sometimes still feel) inadequate as a developer. Every accomplishment I got at work didn’t matter if I couldn’t get the next one. And here I was seeing post after post of people accomplishing things I could only dream of doing, developing applications so complex and cool that I felt I could never touch.

But the more time has passed, the less I view GenAI as a threat than as a tool. There are certain things I will confidently stand for and say GenAI can never replace humans in. Artwork, Film, Movies, Great Writing. All these things require a certain something that GenAI can’t have because it doesn’t work like we do. Someone still has to push the buttons, move the equipment, do the work that AI can’t. And I know some will point to robotics as a avenue to do these things in, but the cost-benefit of robots replacing humans is still way off. It takes way to much time, materials, and money, to produce a machine that: (i) Shorts out if it gets wet. (ii) Relies on a preprogrammed system of commands and systems to interface with our physical world. (iii) Consists of a bunch of subsystems either designed for one specific purpose, or meant to emulate human systems already.

Something that AI can’t do without expensive computation and time is abstract away. We are such good organisms because we can handle the almost infinite complexity of the world around us. I don’t worry about the individual threads of the couch I am sitting on, or the millions of transistors in this computer. They still exist and matter, but humans are able to work with the pieces we see and organize at such a fundamental level to understand a system as complex as our world.