What Huxley's 1961 Interview Taught Me About Algorithms

I watched a 1961 BBC interview in which Aldous Huxley talks with John Morgan. One idea from that conversation has stayed with me: the direction technology develops in can drift away from the life people actually want to live. Working with data for a living, the conversation made me reconsider whether it’s enough to call a technology “good” just because it makes something more efficient.

What came to mind as I watched was recommendation algorithms. They help us pick the next video to watch or the next article to read, but sometimes we end up watching content quite different from what we originally set out to find. Whether the behavior a service counts as a good outcome is also a good experience for the user is a separate question worth asking.

Reading Huxley’s Warning Alongside Today’s Technology

According to BBC broadcast records, this interview aired on July 30, 1961. In it, Huxley shares his thinking on society, technology, and utopia at the time.

What worried Huxley was that the purpose a technology is built for doesn’t necessarily match the direction it expands in once it’s loose in society. A convenience we accept for its own sake can later become a condition people have to adapt their lives around. I’ve come to see the way today’s services arrange our options and nudge us toward the next action through this same lens.

Jacques Ellul’s The Technological Society is worth thinking about alongside this. Ellul analyzed how technologies and organizational methods that raise efficiency spread into every corner of society, crowding out other values along the way. Here, “technology” means something broader than a single machine.

Connecting this discussion to recommendation systems is my interpretation, reading as someone living today. Huxley never described the specific design of reinforcement learning or social media in advance. I want to keep two things separate: the fact that his concerns still give us something to think about, and the question of how today’s technology actually works.

What Does a Recommendation System Consider a “Good” Outcome

Recommendation systems rank content you might be interested in, based on information like your usage history. But what exactly gets predicted and evaluated is a decision each service makes on its own.

For instance, YouTube stated in an official 2021 explanation that it uses not just clicks and watch time but also satisfaction surveys, shares, likes, and dislikes in its recommendations. The reasoning given was that long watch time alone doesn’t necessarily mean that time was valuable to the user — hence the surveys. For news and information content, the company also said it factors in quality and reliability.

So it would oversimplify the actual architecture to claim that every recommendation algorithm is designed purely to maximize time spent. Still, the underlying question — what counts as measuring a “good” outcome — remains unresolved. Clicks, long viewing sessions, and high satisfaction scores each capture only a slice of the user experience.

What I think matters here is checking the service’s goals against the user’s actual purpose, side by side. Someone who searches for a lecture they need and watches it in full, and someone who wanted to stop but kept watching because the next video looked interesting, might rack up identical watch times — yet they deserve very different evaluations. If a system fails to distinguish between these two cases, its metrics can keep improving even as it drifts further from what the user actually wanted.

Why Brave New World Comes to Mind

Huxley’s Brave New World shows how freedom can be constrained even in a society that offers pleasure and stability. What I want to think through here is whether feeling satisfied is enough, on its own, to say we’ve genuinely examined our own choices.

That said, I’m not equating the controlled society in the novel with social media outright. Recommendation features help us discover material we’d struggle to find on our own, and connect us with content from people who share our interests. What I want to ask through this novel is whether we can stay conscious of what we’re choosing, even while enjoying that convenience.

The same question applies to services where we converse with AI. A conversation lasting a long time, feeling a sense of closeness, and actually getting the help we wanted are three different experiences. It’s hard to account for all three just by looking at how long a conversation ran or whether someone came back.

The Word That Stood Out to Me in Island

Huxley’s final novel, published in 1962, Island, is set on the fictional island of Pala. This society draws on both Eastern meditation traditions and Western science, helping its members cultivate their own capacities.

One phrase the birds in the novel keep repeating is “Attention.” I read this as a call to notice what you’re actually experiencing right now. Rather than telling us to abandon technology altogether, it connects to the idea of not losing sight of your own purpose while using it.

The society in the novel is vulnerable to outside political and economic pressures. So I don’t want to read this book and conclude that everything gets solved as long as individuals focus hard enough. I think we need both: examining our own usage habits, and scrutinizing what choices a given service is designed to nudge us toward.

There’s also institutional movement toward scrutinizing how digital services are designed. The European Commission launched a public consultation in July 2025 to prepare a Digital Fairness Act. Under review were things like addictive design, deceptive interface layouts, and personalization that exploits user vulnerabilities. At that point, the process was still at the stage of legislative preparation and public consultation.

Oswarld’s Lens

I’ve often said, in the course of my work, that what isn’t measured is hard to manage. After watching Huxley’s interview, I want to add a clause in front of that: you first have to decide, carefully, what to measure.

I’ve seen companies roll out algorithms to improve user experience and watch satisfaction scores climb. There’s no need to deny that this is a real achievement. But I think we still need to ask whose satisfaction was measured, what exactly was asked, and whether the result actually aligns with what users want over the long run.

What worries me is when this kind of scrutiny stops simply because the metric looks good. Even when a model hits its stated target perfectly, any value that wasn’t built into that target still needs to be checked separately. The very act of measuring efficiency and satisfaction already involves a judgment call.

This brought to mind the design principle of thinking about form in terms of function. It’s not enough for a function to be clearly defined — what matters just as much is who it’s meant to help, and how. Whether a feature is designed to save users time or to keep them consuming more content changes what counts as a good outcome.

So I want to try holding onto the idea of “attention” from The Island while I’m actually using these products — to check what I opened the app to do, and whether what I’m looking at right now actually serves that purpose. It’s not some grand solution; it’s just a way of double-checking my own intent before I let a recommended choice carry me along.

References & Further Reading

• BBC Archive, 1961: Aldous Huxley on the power of TECHNOLOGY! — This is the interview clip I watched. You can check the broadcast date and host in the BBC schedule archive.
• Jacques Ellul, The Technological Society — A book analyzing a society organized around technology and efficiency.
• YouTube, On YouTube’s recommendation system, 2021.9.15. The company’s own explanation of the signals it uses for recommendations and how it manages quality.
• Aldous Huxley, Island — The original was published in 1962.
• Neil Postman, Amusing Ourselves to Death, 1985. A useful companion read connecting Huxley’s and Orwell’s work to popular-culture criticism.
• European Commission, public consultation on the forthcoming Digital Fairness Act, 2025.7.17.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.