Do you want better recommendations online? It takes better AI and the human touch

Spending hours in endless rows side-scrolling Netflix movies or searching forever long lists of similarly rated restaurants on Yelp – that just can’t be the way it’s supposed to work. Part of the entire promise of the Internet is that platforms and services will take the web’s endless supply of everything – the things to watch, read, look at, play with, buy, eat, invest, comment on, listen to, get feelings about – combine with Deep understanding of who you are and what you love, sending you an endless supply of all your favorite things.

When it works, it can feel magical, like the TikTok algorithm that seems to know you more than you know yourself. But this is very rare. All too often, you’re chased online by Amazon ads for products you’ve already purchased, or you’re stuck browsing hundreds of 3.5-star Yelp listings or hundreds of true crime-like podcasts on Spotify only to find something you like. Or you end up watching the desk. second.

Good recommendations seem like a simple enough problem, right? The companies and platforms that work on these personalization machines say it’s a harder problem than it looks. Mostly because humans, you see, are hard to spot. But they also say there is a way to do a better job. And a way you can help.

When the team first started implementing content recommendations similarly to building their platform, they thought the best way to do recommendations was to build a social network. CEO Ian Morris similarly says, “What happens in real life, do you go out to lunch or dinner, and the first thing after ‘How are you, how are the kids’ is that you talk about the things you want to read or that great new show you watched or a podcast? You really need to start listening to him. That’s life!” He felt that on the Internet, those human links and recommendations were being replaced by bad algorithms that were optimized for sharing and growing on actual quality content. He also believed that it could be a source for finding movies, shows, books, and podcasts, all in one place.

Morris remains convinced that this is the right approach. It didn’t take off as quickly as he had hoped, although building a social network from scratch is pretty hard work – and so he started thinking about how to make the platform more useful even for someone who didn’t have a large group of friends using the same method. I hired an editorial team to scour the internet for the best new and most interesting stuff, and at the same time started building a machine learning system that could make automated recommendations.

It likewise collects all the things you want to watch and all the things you think you should watch.
Image: likewise

Now, when you first start using Likewise, it requires you to tell it the things you like. If you want to get movie recommendations, first choose two genres – comedy, drama, and western – and then pick some of your favorites from a curated selection of titles. You can’t access the rest of the app until you choose at least 20. “The payoff is huge,” says Saleem Hamdani, chief technology officer at Likewise. “The more you tell us, the better.” He says people never stop at 20 because it’s fun picking out the things you love. In doing so, the Likewise algorithm tells who you actually are.

Likewise, it uses this information to put you in a “group,” which refers to a group of people with similar tastes to yours. These collections are constantly changing based on what else you watch and rank, and they inform you about everything else and likewise recommend you. “It gives us a starting point to say, how many people in the world are like you, and how many groups can we create?” Hamdani says. The more precise and specific those combinations are, the more accurate they are. I know you love Succession A little useful I know you like Succession, Michael Crichton’s Podcasts Adventure Zone And anything that has Marvel in the title is vastly more useful.

The simplest and most widespread recommendation system, similarly and elsewhere, is known as collaborative filtering. It works by assuming that if you like one thing, and another person likes that thing and also the second thing, then you probably like the second thing too. This is it! It usually involves more data and more people, but that’s the basic idea: if you want to to cut And the other people who loved to cut They really dig old manMaybe you will, too.

One of Morris’ theories is that similarly better recommendations can be made, not just by knowing users better, but simply by having more things to offer them. Netflix, HBO, and Disney will never recommend each other’s catalogs, but similarly (along with apps like Justwatch and Reelgood) they can index them all. “We are not aware of any recommendation engine that looks at things like the social graph or searches across books, podcasts, TV shows, and movies, and allows your preferences and other things to influence each other across these categories,” Morris says.

The simplest way to get better recommendations, almost everyone in this space has told me, is to give apps and platforms more to work with. Many executives have described the ideal personalization process as a collaborative exercise in which you and AI work together to paint an accurate picture of what you really love. Everything you love about Netflix helps the app put you in the right groups; Each filter you select on Yelp makes restaurant recommendations more useful. Dissent and dislike votes are equally useful. Clicks, likes, and even shares can mean a lot of things, but an outspoken endorsement sends a much stronger signal.

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

Oddly enough, though, many platforms have gone the other way, choosing to infer what you want based on what you’re clicking on, slowing down as you scroll, or engaging in some way. It’s based on the desire for a completely frictionless user experience, but from Facebook to YouTube to TikTok, we’ve seen what it can lead to: misinformation, rabbit pits, echo chambers, and problems of all kinds. It also requires collecting amazing amounts of data, grabbing every possible bit of 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 drive toward zero friction. He says, “If you open up your favorite streaming platform and you’re going to watch a movie, I don’t think you want to first answer a 30-question quiz: Hey, what are all your favorite movies? Well, how do you rate them? Who are your favorite actors? I don’t think any Someone who wants to do this work.” Instead, he says, the key is to find the right moments to ask questions. “I have a barber that I’ve been going to for 10 years to get my hair cut,” says Gaffney, for example. “And if you think about this experience every time, it’s a personal one, and I don’t need to tell him when I walk in how I want my haircut because he knows me. But it started with the first conversation: It was a frank conversation, like, ‘Hey, how do you generally like to cut your hair? ?” Making the same kind of dialogue clear, without overusing it, is a major goal of Pinterest.

A side effect of this collaborative process is that it can also provide users with more transparency about what is recommended and why. Almost everyone I spoke to about this story said that this is important in helping people have good experiences online and in creating trust in the things that are recommended. “More and more, I think we want to know: What are the decisions? What are the things that inform some of these algorithms that actually give us content?” says Gaffney.

Confidence is really everything. There’s a default version of Yelp — the Netflix app, the Spotify app, the Kindle app, and dozens of others — and that’s nothing more than a big button. You sit down to watch something, you press the button, and Netflix knows exactly what you’re looking for. Spotify puts out exactly the right song. Scream asks for the exact dish you crave. Everything is personal, automated and delivers one true recommendation at a time. But can you believe that it is enough just to press the button? Akhil Ramesh, president of consumer products at Yelp, doesn’t think so. “I often joke that if God came down in front of me and said, ‘This is the person you’re going to marry, and you’ll never have to waste a second,’ I wouldn’t believe it,” he says. “I’m going to go do an exploration.”

The one true recommendation isn’t just impossible – it’s not really worth pursuing. But this does not mean that things cannot improve. As the services we use improve in our knowledge – and just as importantly, improve in our asking about ourselves – they may be able to narrow the world down to a few options rather than an endless scrolling list. All you have to do is choose your favorite and go. Because, really, there he is No correct answer. There is only the one you chose.

Leave a Comment

Your email address will not be published.

beautiful home decor Williams vs Radokano match postponed Brian Kelly, LSU player Myles Brennan, retires from football Black Adam and Stripe are seemingly heading to MultiVersus Bryce Dallas Howard claims she received payment.
beautiful home decor Williams vs Radokano match postponed Brian Kelly, LSU player Myles Brennan, retires from football Black Adam and Stripe are seemingly heading to MultiVersus Bryce Dallas Howard claims she received payment.