Ask HN: How many AI chat interfaces do we need?

Many years ago I felt like chat would become the dominant interface, but I never really had this realisation about llms or natural language use for commands. I just thought it would be a lot like a thousand commands in something like a personal slack. But now we have this AI base chat interface and I'm just wondering how many do we need? ChatGPT, Gemini, etc. I'm guessing creating another one without some point of differentiation isn't useful. So what does the future look like? Do we just converge to a few of these? Will we see others emerge? Or do we see vertical specific niches?

Curious to know the thoughts here