I Don’t Need 25 Recommendations. I Need One Good One.
I have spent an embarrassing amount of time trying to find something to watch. Not watching something. Finding something.
I finish a series I really enjoyed, open Google, and type something like “shows to watch if you liked Only Murders in the Building.”
And Google delivers.
“25 Shows to Watch If You Loved Only Murders in the Building.”
Perfect.
Except I don’t need 25 recommendations. I need one good one.
So I open the article. I’ve already watched six of them. Three are on a streaming service I don’t have. A couple are much darker than what I’m looking for. One apparently made the list because it also takes place in New York. And somewhere around recommendation number 14, I forget what I was looking for and go back to scrolling through Hulu.
This is one of those ordinary little tasks where I’ve realized AI has quietly changed how I do things.
Because Google knows what’s popular. AI can learn what I like.
Search starts with what you’re looking for
Google is really good when I know what I’m looking for.
If I want to know when the next season of a show comes out, who wrote a book, or where a particular hotel is located, search makes perfect sense.
But recommendations are different. Usually I don’t know exactly what I’m looking for. That’s the whole reason I’m asking. I may know that I want something funny but not silly. I like mysteries, but I don’t want anything scary. I want characters I care about. I don’t want to commit to 14 seasons. I’ve already watched Only Murders in the Building, Veronica Mars, The Good Wife, Elsbeth, Red Oaks, and half the other shows that will appear on somebody’s “Top 25” list. And tonight I’m tired, so please don’t give me something that requires a flowchart and a Reddit community to understand the first episode.
Try putting all of that into a Google search box. With an AI assistant, I can just say it. That’s the difference.
The conversation becomes the search
The first recommendation doesn’t even have to be right.
Maybe the AI recommends something, and I say, “I’ve already seen it.”
Or, “Nope. Too dark.”
Or, “I loved that one. More like that.”
Or even, “That sounds good, but I don’t have Hulu.”
Now it knows something it didn’t know before. We aren’t starting the search over. We’re narrowing it.
That’s what makes recommendations through an AI assistant feel so different from scrolling through search results or “Best Of” lists. I don’t have to translate what I want into the perfect keywords. I can describe what I want the same way I would describe it to another person.
And I can be ridiculously specific.
“I want somewhere warm to go in February. I’d prefer a direct flight. I don’t want to rent a car. There needs to be plenty for a preteen to do, but I would also like somewhere pretty where I can sit while she does those things. Nobody wants to hike up a mountain. We have four days. And don’t say Disney.”
That’s not a search query. That’s a conversation. And that’s exactly why it works.
It can tell me why
This may be the part I like most.
A good recommendation shouldn’t just tell me what. It should tell me why.
If an AI assistant recommends a television show because I liked the character relationships in one show, the mystery in another, and the humor in a third, I can decide whether its reasoning makes sense.
The same thing works with books.
Don’t just give me a bestseller because everybody else liked it. Tell me, “I’m recommending this because you liked the family dynamics in this book, the writing style in that one, and you said you don’t want something with a completely depressing ending.”
Get Tracy Buckner’s stories in your inbox
Join Medium for free to get updates from this writer.
Now I have something useful.
The recommendation isn’t based only on what millions of other people clicked, watched, bought, or rated. It’s based on what I’ve told the AI matters to me.
This works for a lot more than television
Books are an obvious one. So are movies. But think about vacations. Restaurants. Gifts. Recipes. Activities with kids. Even deciding what to do on a Saturday afternoon.
The traditional search process often starts with what’s available or what’s popular.
“10 Best Family Vacation Destinations.”
“20 Best Books of 2026.”
“15 Restaurants You Have to Try.”
“30 Perfect Gifts for a 10-Year-Old.”
Those lists aren’t necessarily bad. Sometimes they’re really useful.
But they were written for everybody. And I’m not everybody. Neither are you. An AI assistant lets me start with what makes my situation different.
My budget.
My schedule.
What I’ve already tried.
What I loved.
What I hated.
Who I’m traveling with.
How much energy I have.
What I’m absolutely not willing to do.
Sometimes the constraints that would make a Google search ridiculous are exactly what makes an AI recommendation useful.
There’s still a catch
Of course there is. AI can be wrong.
It can recommend a restaurant that closed six months ago. It can tell me a movie is available on a streaming service when it isn’t anymore. It can get a hotel location wrong or confidently describe a book based on incorrect information.
So I don’t hand over my credit card because an AI assistant says, “Trust me, you’re going to love Aruba.”
AI is excellent at helping me narrow the possibilities.
But when the facts matter, I verify them.
If we’re planning a vacation, I’ll check the actual airline schedule. I’ll look at the hotel’s website. I’ll verify prices, hours, availability, and anything else that could have changed.
That’s where human judgment still matters. The AI helps me get from thousands of possibilities to a handful worth considering. I decide what happens next.
Search isn’t going away
I don’t think AI assistants replace search. I use both.
Search is often better when I need to find a fact, a website, a product, or something specific. But when I don’t know exactly what I’m looking for yet? That’s where conversation has become much more useful to me.
I can explain what I like. I can add constraints. I can reject an idea without starting over. I can ask why something was recommended. I can change my mind halfway through.
In other words, I can behave like a person instead of a collection of keywords. And maybe that’s the real shift. Search starts with the thing I’m trying to find. AI can start with me.
Because most of the time, I really don’t need 25 recommendations. I just need one good one.
