Michael Rowe

Trying to figure out better ways to help people learn.

Perplexity AI is a great service when accuracy is important

I’ve been using Perplexity on and off for almost as long as it’s been available. I’ve always liked the way it provides sources for it’s responses, but lately it seems to be really good. Like, really good. The sources are almost always places I recognise and trust (when it’s a topic I know about) and the in-line citations are great.

Perplexity AI user interface.

Another feature I really like is the Related section, where Perplexity suggests follow up questions and I often find myself going down some interesting rabbit-holes.

Perplexity AI related section.

I also like the fact that it moves past threads (conversations) into a Library section, and does the same with the Discover section (something I hate about other services is how much they clutter the landing page with suggestions of rubbish prompts). You can also add conversations to Collections.

And it shows it’s process for narrowing down the subset of potential sources. For me, this helps to trust it’s reasoning.

Perplexity AI showing it's sources and reasoning process.

The free version is limited; you have a cap on uploading 3 documents per day, and you only have a few requests that can use Copilot, but if I’m just asking regular questions like I used to go to Search for, then Perplexity is increasingly my go-to option.

The latest version of Bard (the one that uses Gemini) has also impressed me the with quality and accuracy of it’s responses.

Note: All the usual caveats apply; generative AI makes things and, in general, you should not trust it for anything at all. Except, increasingly often, you can. But you shouldn’t.


Discover more from Michael Rowe

Subscribe to get the latest posts to your email.