Dev·stopian

With the acceleration of generative AI at the end of 2022, the future looked developer-dystopian (dev·stopian). If AI could produce code in a wide range of domains, then what would my role be? Some of the code it was producing took me years to master. Within days, I shifted my attention to augmenting my capabilities. That felt like a gasp of air. I turned my attention to what I was now able to produce. I began pressing AI harder and harder to squeeze every last drop of knowledge that I could. I began to level up my skills at an accelerating rate; I was no longer relegated to searching the web for hours (or years in some cases) before making progress. I took on challenges I thought were years in the future.

GenAI pushed us all to be readers. Sure, we lay down our prompts with ease, but we are inundated with information to read. Read, read, read, and more reading! It feels like my job turned into reading. I used to write...a lot. Some boilerplate here, function there, struct over there, save, build, verify, repeat. It felt so natural and productive. Now, I think, prompt, read+, build, verify. In comparison, I barely even write.

But wait, why write at all? AI can produce the quantity and frankly, the quality way better than me. I need to write; I need to take the time to reflect and synthesize the things that I have been learning. All of this reading must be followed up with writing, or I'll lose it (ambiguity intended). I found the best way to learn something is to try and explain it.

Now, I'm going to push back on the idea of quality output of GenAI. In some aspects, this is certainly true. However, a statistically smooth response (predictable outcomes for next tokens) can help inform what "the Internet" might say about a topic, but writing needs to be rough. We need content creators to keep being rough. How else can we maintain the edge?