Average work isn't good enough anymore
I've been reading a lot about how "taste" is more valuable than ever - that AI will be best leveraged by those with good taste. I largely agree with this sentiment, but it only hit me recently that not many people are talking about the opposite: the reality that those who aim for "average" decisions or output will find themselves increasingly commoditized by AI.
Historically, it could be argued that the "average employee" has been incredibly important to a company. After all, "average" decisions tend to be solid decisions, and the path you want to take most of the time, e.g. the variable names, tech stack, frameworks, etc. used to build a feature might all be a series of "average" decisions. You really want to save the risky decisions for when they're most worthwhile.
However, as I'm sure we've all seen with the rise of AI in coding, AI is incredibly good at making average decisions by default. This makes sense - conventional decisions are well represented in its pre-training data (the internet), score well across a variety or RL scenarios post-training, and tend to be the most agreeable based on human feedback as well.
This isn't to say that a majority of people should find themselves out of a job due to how a bell curve is shaped. Rather, the average that is being commoditized away are average decisions, rather than a notion of "an average person" (if any such thing even exists).
The upside is that we're most valuable in spending our time making critical decisions that aren't just a rinse & repeat of what's been done before. This isn't to say we should go out and pick the riskiest database to build on, but rather spend our time focused on evaluating when does it make sense to buck the norm and make the unintuitive decision.
AI frees us from being buried in routine decisions, and that's an exciting future for those that love thinking beyond the average.