Is AI Better at Picking Your Next Show Than You Are?

With AI becoming more prominent, it’s no surprise that it may be choosing what you watch next on streaming. Instead of spending a chunk of your Friday evening searching for something to watch, you can simply choose from the AI recommendations. Can we be sure that these recommendations are perfect for us? Let’s dive into just how AI optimizes its recommendations and see how accurate it is to match our interests.

When scrolling AI recommendations on a streaming platform, you might assume it’s solely based on your watch history. Interestingly, it’s a lot more complex than that.

Out of all the streaming platforms, Netflix has the most advanced AI recommendations, thanks to its Foundation Model. As detailed in Litslink, “It treats your viewing history like a sentence: every click, pause, scroll, and rewatch becomes a ‘token,’ and the model reads that sequence to predict what you want next.” Essentially, it dives deep into your viewing, clicking, and scrolling habits.

Here is an idea of what the Foundation Model tracks:

• All previously viewed content, including genre, cast, and crew
• Rewatch frequency
• Viewing consistency (binge-watching vs. sporadic viewing)
• Completion speed
• Content ratings
• Show drop-off/abandonment

Let’s face it — at some point, the thumbnail alone drew you into a show or deterred you. Recommendations are more than a curated list of content in a category such as “You may also like.” The AI goes through all frames of a movie and selects the most alluring thumbnail that will pique your interest.

Netflix even used AI to generate trailers when promoting “House of Cards.” The AI generated 10 different trailers, each tailored to attract different audiences — some saw their favorite actors while others saw appealing plotlines. With Netflix being the pioneer for AI recommendations, all other streaming platforms are following suit. And although they’re not yet as advanced, there’s a growing reliance on algorithms.

What does this mean for the viewer? AI knows when I’m pausing a Marvel movie to find all the Easter eggs and when I stopped watching “Bob’s Burgers” after four seasons. My first thought was that this should be ideal — I would obviously want my streaming platform to recommend things I like and hide everything else. However, this is a mixed blessing.

As Tom’s Guide writer Dave Parrack highlights, AI recommendations “are all based on your previous interactions, meaning you’re stuck in a dwindling spiral of content.” Ultimately, even with its deep knowledge of our viewing habits, AI can only recommend more of what we've already watched, not what we might actually want to discover next.

Parrack tested Netflix’s algorithm by only watching AI recommendations for a week. At first, he enjoyed these recommendations. However, Netflix began to recommend clones of the same content. The overall conclusion: “Netflix's algorithmic recommendations are remarkably useful for removing friction from the viewing experience … However, that convenience comes at the direct expense of variety. The recommendation engine is a great guide, but a terrible explorer — the latter still requires a human touch.” Although AI is advanced in curating recommendations for you, it only knows what you’ve watched, not your potential interests.

The purpose of these recommendations is to ensure that users are always watching — the quicker you start watching, the less likely you will cancel your subscription. Ironically, these repetitive AI recommendations may contribute to overall streaming fatigue. Users are sick of price hikes, keeping track of multiple services, and the crackdown on password-sharing. As a result, people are considering switching to YouTube, ad-supported subscriptions, and even a digital antenna. Imagine the good old days free of an exhaustive library, monthly fees, and AI interference.

I typically don’t pay any mind to the recommendations on my streaming platforms since I know exactly what I want to watch. Checking out “Today’s Top Picks for You” on Netflix, I find recommendations that are both predictable and surprising:

“Spider-Man: Homecoming” and “Spider-Man: Into the Spider-Verse” are on the top of my list despite the fact that I watched them recently. I assume these movies are here because many users with my similar tastes are currently watching Spider-Man films due to the current hype of the latest release: “Spider-Man: Brand New Day.”

“Pop Culture Jeopardy” is probably the one recommendation that I’ve never heard of that somewhat piqued my interest. No doubt, this was recommended because of the pop culture and cult classics I watched.

“The Fragrant Flower Blooms with Dignity” is an anime recommendation that surprised me since I’m currently watching it right now. With Netflix’s growing library of anime, it’s quite confusing as to why the AI chose the one that is on my Continue Watching list.

“Jeff Arcuri: Nice to Meet You” is a stand-up comedy special that probably baffles me the most. Not only have I never heard of this comedian, but I haven't watched a comedy special in over two years; I would think the AI would’ve known this.

Perhaps the laziest recommendations I have are from the section titled “Because you watched The Fragrant Flower Blooms with Dignity.” With this romance anime, I would expect a list of other romantic content. Instead, the list is just a random spattering of anime spanning multiple genres, such as action and horror. This recommendation is simply “If you like anime, then you will enjoy anime.”

Overall, most of the AI recommendations were either content I'd already seen or content I had no interest in, aside from one title that mildly interested me. It’s safe to say that I won’t be relying on AI to choose my next watch for me.

Despite AI recommendations being generally unfavorable, the practice isn’t all bad. If you know what genres you love and won’t grow weary of the same tropes repackaged on a loop, you will always have content to watch. If you want to actively fight against AI without denouncing streaming, you can commit to subscription nomadism. Otherwise, you can be like me and create your own curated watchlist — sometimes you just need to rely on your own recommendations.

Found this useful? Follow Reviews.org on Instagram, Facebook, LinkedIn, or catch us on YouTube. 

You can also add us as a preferred Google search source for the latest in tech and home services.