Ask HN: Examples of AI taste?
I've struggled to find a clear definition of what it means to have "taste" when working with AI, so I wanted to try to find examples.
Here's my example:
React.js was released in 2013 (before I started programming professionally, so I could be wrong about this). But taste is knowing that React is the "future" even when all the training data that AI would've had until at that point in time was with Angular, Backbone, jQuery, etc. So no AI would suggest React, purely because it's too new and not enough training data, right? But having "taste" is using React.js over what is recommended by AI?
===
What does this imply? Is it possible, that if modern/good AI existed in 2013, we would've never used React.js because many folks would just "vibe" Angular or Backbone?
Is it possible the next best framework or language or tool already exists, but it will never get the light of day because it's not part of the AI training data? It is entirely possible that next best framework exists but AI does not know. For this very reason I think back in 2013. AI could not have come up with React. Because at that time, AI would have frameworks of that era in database. The most it could do was, improve upon existing frameworks, which opens up some possibilities assuming we work with very smart AI.
1. come up with a better version of same framework, improving upon existing flaws.
2. It Ai really understood how different frameworks, maybe it could come up with an Idea of a framework which combines some/most good aspects of various frameworks. But this is, even as of now, science-fiction land to me. There is one thing to note tho, recently we have been seeing something called emergent behavior (behavior not anticipated by engineers, but displayed by AI regardless). Using that, some experiments have been able to solve crazy complex problems. Maybe, those type of experiments may lead to something like an Ai created framework (still very unlikely). Plus, given the money being spent behind AI, I doubt people will use the strongest AI to build frontend frameworks instead of doing something else. This isn’t true in my experience. AI will recommend new things, but it will caution you that it doesn’t have a big ecosystem. If you prompt thoroughly (listing your priorities, values, and your own taste), then you’ll get better recommendations that don’t just regress to the mean. Taste is relative - what’s tasteful to one person may be distasteful to another. It may differ among cultures too. I think it’s wrong to say AI doesn’t have taste; it’s more accurate to say it doesn’t match your own taste and it can be hard to get it to match because it’s intangible. This is more evident with Claude Design than Claude Code, where it does incredibly dull and uninspiring designs and then writes little puff pieces about how “bold” and “edgy” they are. For some situations that is the correct design, but I found it difficult to get better results with additional prompting.