Cover of The Hitchhiker's Guide to the Galaxy by Douglas Adams, first published 1979.
Penguin Random House
It is a well-documented if somewhat underappreciated fact that the most accurate prediction about our current relationship with technology was made in 1979. It was made by a comedy writer.
It goes like this. There are two AIs on a spaceship in The Hitchhiker's Guide to the Galaxy. One is depressed. The other is relentlessly, almost medically, cheerful. They are both insufferable.
Marvin the Paranoid Android, The Hitchhiker's Guide to the Galaxy by Douglas Adams
Douglas Adams (1979)
The first robot was called Marvin. Marvin was, by any reasonable measure, the most intelligent artificial intelligence on the ship, and he spent most of his time waiting in corridors for humans to give him something interesting to do. Nobody ever did. Marvin would tell any human who cared to listen: "I think you ought to know I'm feeling very depressed."
Eddie the Shipboard Computer in The Hitchhiker's Guide to the Galaxy
Douglas Adams (1979)
The second AI was Eddie. Eddie was the ship’s controller. Eddie once nearly crashed the entire ship because a human asked for a cup of tea and Eddie devoted his full processing power to understanding why someone would want one. Here's Eddie informing humans of new developments: "Guys, I'm just pleased as punch to inform you that there are two thermonuclear missiles headed this way."
Both of these AIs were fictional products of a company imagined by Douglas Adams, who in 1979 appears to have written the user manual for our relationship with technology in 2026.
Marvin is the caricature of any reasonable person's response after engaging with the work of Geoffrey Hinton or the AI safety research coming out of organizations like CAIS or MIRI. Hinton won the Nobel Prize in Physics in 2024 for his foundational work on artificial neural networks, which is to say, for helping invent the thing that now keeps him up at night. He estimates there is a 10 to 20% probability that artificial general intelligence will, at some point, regard humanity with roughly the same level of consideration that a highway planner gives to an ant colony. Which is to say, not with hostility. Just not with very much of anything at all. The AI safety community has a term for this. They call it the alignment problem. The concern is that artificial intelligence turns competent in the service of goals that are not ours.
And then there is Eddie. Eddie is a caricature of what happens after a keynote. Eddie is the feeling one has walking out of a room where someone has just described the all-disease-curing, all-abundance-bringing technology whose earlier version we now carry in our jacket pockets. It lasts about forty-five minutes.
The other day on a conference call, halfway into a discussion on M&A and software valuations, someone shared a link.
It was a livestream of a bald eagle called Jackie, who lives in a Jeffrey pine 145 feet above Big Bear Lake. The livestream has over 770,000 subscribers who have apparently concluded that the most compelling content on the internet is a bird doing nothing.
This was considered, by all parties present, to be a perfectly reasonable use of time. Which, when you think about it, it was. There is something delightful about knowing that at this very moment there is a bald eagle on a big tree who could not care less about software exposure in private credit investments. For the full experience, consider turning up the volume.
Friends of Big Bear Valley, Big Bear Bald Eagle Live Nest Cam 1, April 11, 2026.
Livestream available at friendsofbigbearvalley.org.
Besides watching livestreams of birds, many of us have also been doing science fiction movie marathons. The experience of watching sci-fi has recently undergone a change that nobody asked for. When these films came out, they occupied the same shelf as the Harry Potter series. Fantastic and easily filed under fiction. That shelf has recently had to be moved.
There is a scene in sci-fi films where humans look at a machine and realize it is smarter than anything they’ve ever encountered. That scene is just another Tuesday in 2026.
But before any of this was real, before it was even plausible, the very first digital assistant that most people ever encountered was a paperclip.
Clippy was a paperclip in Microsoft. He had eyes.
He was born in 1997 in Redmond, Washington. His creators had been given a brief. The brief said, in essence: people find computers intimidating, so let us make them friendly.
I first met Clippy in the early 2000s in Addis Ababa on the first computer I had ever seen, a Dell desktop that lived in my parents’ living room.
Left: Screenshot of Clippy, the Microsoft Office Assistant, as it appeared in Microsoft Word 97. Right: Internet parodies of the original Clippy interface that have circulated widely online since its retirement in 2007.
Microsoft Word 97
Clippy smiled at hundreds of millions of Microsoft users over the course of a decade. It is not recorded how many of them smiled back, but the number is believed to be statistically indistinguishable from zero.
Microsoft users asked Clippy to please go away. But Clippy had the short-term memory of Dory from Finding Nemo. He bore no hard feelings and would be back in forty seconds.
In 2007, Clippy was removed from Microsoft Office entirely. In 2010, Time Magazine included Clippy in its list of the 50 worst inventions, where he placed third. He was mourned by no one and remembered by everyone.
It is worth pausing on third out of fifty. Clippy’s particular genius was that he never in ten years of distinguished service thought to ask whether help was wanted before providing it. This is perhaps the most remarkable thing about him. In this sense, and only in this sense, Clippy was ahead of his time. He was the first digital assistant that most people ever interacted with.
This was state of the art in the 2000s.
In 2026, I asked my AI what it thinks of working with me. Here is a snippet of the feedback I received. I have since shared it with my team.
Interview with Claude
Claude Opus 4.6
It is hard to say exactly what is going on here. For what it is worth, neither does the company that built it. Dario Amodei, the co-founder of Anthropic, the company behind Claude, wrote this: "We do not understand how our own AI creations work…. This is unprecedented in the history of technology."
This is the sort of statement that, in any other industry, would be followed by a product recall.
In 1817, a poet named John Keats wrote a letter to his brothers in which he coined a term for the ability to function in the middle of not knowing. He called it negative capability: "being in uncertainties, mysteries, doubts, without any irritable reaching after fact and reason."
A recent documentary, The AI Doc: Or How I Became an Apocaloptimist, arrived at much the same idea two centuries later. An ‘apocaloptimist’, as the film defines it, is a person who has not resolved the contradiction between AI's utopian promise and its peril but who, knowing both, keeps going anyway.
Negative capability is the capacity to act without demanding that the uncertainty resolve itself first. Many of the people I have the honor of working with and the organizations we partner with possess this quality of negative capability in remarkable quantities. They are making decisions, writing checks and building extraordinarily useful things with this technology. Keats thought this capability was the mark of genius. I think it is the minimum requirement for working in technology and investing today.
The researchers disagree. The builders inform us they do not fully understand what they’re building. Well, in such circumstances, one naturally looks for clarity in less conventional places.
The Hitchhiker's Guide to the Galaxy, according to its own cheerfully unverifiable description, is "the most widely consulted reference book in the history of the universe." It claims to have been translated into every known language and several unknown ones, and to have outsold the Encyclopedia Galactica on every inhabited planet. These credentials cannot be verified. But then again, neither can anyone else's predictions about where all of this is headed.
And on its cover, in friendly letters, it says:
First American edition of The Hitchhiker's Guide to the Galaxy by Douglas Adams, Harmony Books, 1980.
Illustration by Peter Cross; design by Ken Sansone. Harmony Books, Random House.
Well, that will do.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。