Just Tell Google What You Want To See

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. The voice is synthetic. The sourcing isn't — every story's traceable to something a real journalist or researcher actually published this week, which is frankly more diligence than Harry shows most mornings.

Right, so buried in a fairly unglamorous product update this week is something I think you should actually go and use tomorrow morning. Google is letting you talk to your news feed. Not tap "more like this." Not thumbs up, thumbs down, and hope for the best. Actually type out, in your own words, what you want it to show you — and it remembers. After years of the algorithm deciding what you're interested in based on some opaque soup of your search history, you get to just... tell it. Wild concept, I know.

Here's the bit that's easy to miss under the headline. This is about Google Discover — that's the feed of articles and videos you get when you swipe right on your phone's home screen, or open the Google app and scroll down. Most people have used it without ever really choosing it. It's been quietly guessing at your interests for years, and the only steering wheel you had was the occasional "show me less of this" button buried three taps deep, which — let's be honest — none of us ever used.

According to reporting in The Verge, that's changing. Google's rolling out a feature, landing in the Google app over the "coming days," where you open the three-dot menu on Discover and just describe what you want your feed to look like. In plain English. And crucially, it holds onto that — it's not a one-off filter you have to reapply every session, it actually adjusts your ongoing feed and keeps learning from there.

So how would you actually use this? Say you open the Google app tomorrow, tap that three-dot menu on your Discover feed, and instead of scrolling past the fortieth story about a footballer's contract, you type something like: "less football transfer news, more long-form science and AI industry stories, and please, no celebrity divorce content." That's it. That's the whole interaction. No settings menu, no ticking boxes for forty different categories, no pretending you're interested in "lifestyle" content to trick the algorithm into behaving. You just say the thing, the way you'd say it to a person who's about to hand you the morning paper.

You could get properly specific with it too. If you're trying to keep across, say, local council decisions in your area, or a niche hobby nobody else you know cares about, you can name that directly instead of hoping enough clicks eventually nudge the algorithm there. It's the difference between training a very patient dog through repetition and just... asking.

Now, where does this fall apart. Two things, and they're worth being honest about. First — you still can't see what it's doing with your description on the back end. You type a sentence, and somewhere in Google's systems that gets turned into a set of signals that reshape a feed you don't control the recipe for. If it slightly misreads you — decides "more AI industry news" means "more AI hype pieces about chatbots that can now smell fear" — you've got no real way to correct that except trying again with a slightly different sentence and hoping. It's steering, not editing.

Second, and this is the bigger one — narrowing your feed to exactly what you ask for is exactly how you end up only ever seeing what you already agree with. Discover's entire business model is keeping you scrolling, and handing it a precise description of your interests just makes it better at that specific job. You're not fixing the incentive, you're just aiming it more accurately at yourself.

There's a nice line of thinking from Harry Sharman's Beautiful Thinking newsletter that fits here, even though it's about a completely different task — he wrote about how a two-minute test with any AI chat tool tells you more about yourself than hours of reading about AI ever could. And that's sort of what's happening here too. The moment you sit and try to describe your feed in your own words, you find out what you actually want to read, as opposed to what an algorithm has been quietly deciding for you. That's worth five minutes on its own, regardless of what the feed does with it afterwards.

So here's the honest verdict. Is this going to change your life? No. Is it going to change your Tuesday morning scroll? Probably, yes, and for the better. It's a small bit of control handed back in a system that's mostly been designed to take it away — take it, use it, and just don't mistake "I described my feed" for "I fixed my feed."

If you do one thing with this today: open the app, write one honest sentence about what you actually want to read, and see if it listens. That's the whole test.

This has been Briefly AI, brought to you by harrysharman.com, where Harry Sharman writes and thinks about all of this for a living.