The paradox of UX personalization

There have been many articles that rail against Spotify and the way it recommends more of the same music — never something truly surprising or interesting. This one broadens the idea to all algorithmic platforms, and I think it encapsulates the pitfalls well:

As we engage, algorithms gradually narrow the selection to match our tastes. While this personalization enhances convenience, it also limits exposure, reinforcing familiar choices rather than expanding our horizons.

The problem statement is sound. The author then goes on to offer a rallying cry to designers:

If UX designers want to preserve the richness of discovery, we must rethink how personalization is implemented. Rather than functioning as a closed-loop system that continuously refines itself based on past behaviors, personalization should integrate opportunities for users to encounter the unfamiliar.

A call for the value of randomness? Perhaps. Think about the chance encounters that changed your life. Was some chance and randomness involved in the friends you have had the longest? The music that first made you think “Oh wow, I like this, I didn’t know this existed”?

As we move towards AI-driven UX-systems that present a different UI to different users based on how they use a tool, what their job roles are, how they prefer to learn, etc. will we continue to reinforce the habits and patterns that prevent humans from learning new things? Finding new preferences? Expanding beyond their bubbles?