Want better online recommendations? It needs better AI and a human touch

Spending hours on endless rows of side-scrolling Netflix movies or browsing endless lists of identically rated restaurants on Yelp — that can’t be the way it’s supposed to work. Part of the whole promise of the internet is that platforms and services would take up the web’s endless supply of everything – things to watch, read, watch, play with, buy, eat, invest, comment on, listen to or having feelings about – combine that with a deep understanding of who you are and what you love, and returns an endless supply of all your favorite things.

When it works, it can feel like magic, like the TikTok algorithm that seems to know you better than you know yourself. But it is quite rare. More often than not, you’re chased around the internet by Amazon ads for products you’ve already purchased, or you’re stuck scrolling through hundreds of 3.5-star Yelp ads or a hundred true-crime podcasts on Spotify just to find something you like. Or do you end up watching Office. Still.

Good recommendations seem like a pretty simple problem, right? Companies and platforms working on these personalization machines say it’s a harder problem than it looks. Mostly because humans, you see, are hard to understand. But they also say there is a way to do better. And a way you can help.

When the Content Recommendations app team started building their platform, they thought the best way to make recommendations was to create a social network. “What happens in real life,” says Ian Morris, CEO of Ilike, “is you go out to lunch or dinner, and the first thing after the ‘how are you, how are the kids,’ it’s “Is that you’re talking about things you’ve read or that awesome new show you’ve been watching or a podcast you really need to start listening to. That’s life!” Online, he felt, those human connections and recommendations had been replaced by poor algorithms optimized for engagement and growth over real-world content. He thought the same could be a resource for finding movies, shows, books, and podcasts, all in one place.

Morris is still convinced that was the right approach. It didn’t take off as quickly as he had hoped – building a social network from scratch is very hard work – so he started thinking about how to make the platform more useful even for those who didn’t have a large group of friends using the same thing. He hired an editorial team to scour the internet for the best and most interesting new releases and simultaneously began building a machine learning system capable of making automated recommendations.

Likewise, collects all the things you want to watch and all the things he thinks you should watch.
Picture: Similarly

Now, when you first start using the Like app, you need to tell it about the things you like. If you want movie recommendations, you should pick a few genres first – comedy, drama, western – and then choose some of your favorites from a curated set of titles. You can’t access the rest of the app until you’ve selected at least 20. “The payoff is huge,” says Salim Hemdani, CTO of Alike. “The more you tell us, the better.” He says people never stop at 20 because it’s just fun to choose things you like. And in doing so, you are telling Similar’s algorithm who you really are.

In the world of referrals, you are the one you group up with

Likewise, uses this information to put you into a “cluster”, which refers to a group of people with similar tastes to yours. These clusters constantly change based on what else you watch and rate, and they inform everything you recommend as well. “It gives us a starting point to say how many people are like you in the world and how many clusters can we create?” said Hemdani. The more granular and specific these clusters are, the more accurate they can be. Know that you love Succession is slightly useful; knowing that you love Succession, novels by Michael Crichton, the podcast The Adventure Zone, and anything with Marvel in the title is much more useful.

The simplest and most popular recommendation system, on Similar and elsewhere, is known as collaborative filtering. It works assuming that if you like something, and someone else likes that thing and also a second thing, you’ll probably like the second thing too. That’s it! This usually involves more data and more people, but that’s the central idea: if you like Breakup and other people who liked Breakup really dig The old manyou probably will too.

One of Morris’ theories is that Like can provide better recommendations, not only by knowing users better, but simply by having more things to offer them. Netflix, HBO, and Disney will never recommend each other, but likewise (with apps like Justwatch and Reelgood) can index them all. “We don’t know of any recommendation engines that look at things like the social graph or look through books, podcasts, TV shows, movies,” Morris says, “and let your preferences and stuff get sorted out. influence each other in these categories.”

The easiest way to get better recommendations, almost everyone in this space told me, is to give apps and platforms more work. Several executives have described the ideal personalization process as a collaborative exercise in which you and the AI ​​work together to paint an accurate picture of what you actually like. Everything you love about Netflix helps the app place you in the right clusters; every filter you tick on Yelp makes restaurant recommendations more useful. Downvotes and dislikes are equally useful. Clicks, likes, and even engagement can mean a lot of things, but explicit approval sends a much stronger signal.

Screenshot of a Pinterest search for

Pinterest has embraced personalization as a collaborative process with users.
Image: Pinterest

Curiously, however, many platforms have gone the other way, choosing to infer what you like based on what you click or dwell on when scrolling or engaging in some way or other. ‘another one. It’s based on a desire for a completely frictionless user experience, but from Facebook to YouTube to TikTok, we’ve seen what that can lead to: misinformation, rabbit holes, echo chambers, problems of all kinds. It also requires collecting astonishing amounts of data, capturing all possible information about you and your habits in case some of it is useful.

Naveen Gavini, senior vice president of product at Pinterest, says he understands the push toward fluidity. “If you opened up your favorite streaming content platform and went to watch a movie,” he says, “I don’t think you wanted to take a 30-question quiz first: Hey, what are all your favorite movies? Ok, how would you rate them? Who are your favorite actors? I don’t think anyone wants to go through this job. Instead, he says, the key is finding the right times to ask questions. “I have a hairdresser I’ve been going to for 10 years who cuts my hair,” Gavini says as an example. “And if you think about this experience every time, it’s a personalized experience, and I don’t have to tell him when I walk in how I want my haircut because he knows me. But everything has started with this first conversation: it was an explicit conversation, like, “Hey, so, how do you usually like your haircut?”

Guessing what you like based on your actions is much harder than just… asking yourself what you like

A side effect of this collaborative process is that it can also offer users more transparency about what is being recommended to them and why. Almost everyone I spoke to for this story said it was important both for helping people have great online experiences and for building trust in the things recommended. “More and more,” says Gavini, “I think we want to know: what are the decisions? What are the things that inform some of these algorithms that actually deliver content to us? »

Confidence is everything, really. There’s a hypothetical version of the Yelp app — and the Netflix app, Spotify app, Kindle app, and dozens more — that’s nothing more than a big button. You sit down to watch something, hit the button, and Netflix knows exactly what you’re looking for. Spotify puts exactly the right song. Yelp orders exactly the dish you want. Everything is personalized and automated and delivers the One True Recommendation every time. But would you believe it enough to just push the button? Akhil Ramesh, head of consumer products at Yelp, doesn’t think so. “I often joke that if God landed in front of me and said, ‘This is the person you’re going to marry, and you never have to waste a second,’ I wouldn’t believe it for a second,” he says. . “I would go explore.”

The only real recommendation isn’t simply impossible – it’s not even really worth pursuing. But that doesn’t mean things can’t get better. As the services we use get better at getting to know us — and, just as importantly, get better at asking us about ourselves — they might be able to shrink the world down to a handful of people. options instead of an endless scrolling list. All you have to do is pick your favorite and go. Because, really, there is no good answer. There is only the one you have chosen.

About Jean R. Manzer

Check Also

Book recommendations by Paulina Porizkova

Welcome to Lifetime, The books section of ELLE.com, in which the authors share their most …