What’s the best superpower? No, it’s not super strength. No, it’s not flying. No, it’s not invisibility. The best superpower is the ability to edit your own mind.

I highly recommend Wait But Why’s “The AI Revolution: The Road to Superintelligence” and “The AI Revolution: Our Immortality or Extinction”. In this two-part series, blogger Tim Urban explores the impact of Artificial Intelligences in our present and near future. The stuff about how society would or wouldn’t function under a hyper-intelligent overlord is fascinating, but the real interesting part is what he says about how AIs get smarter. Basically, he envisions that we will reach AGI, or Artificial General Intelligence (an artificial intelligence that can perform many unrelated tasks and is smart outside of a specific domain) by programming a computer to program itself to be smarter. But what happens when we reach AGI?

An AI system at a certain level—let’s say human village idiot—is programmed with the goal of improving its own intelligence. Once it does, it’s smarter—maybe at this point it’s at Einstein’s level—so now when it works to improve its intelligence, with an Einstein-level intellect, it has an easier time and it can make bigger leaps. These leaps make it much smarter than any human, allowing it to make even bigger leaps. As the leaps grow larger and happen more rapidly, the AGI soars upwards in intelligence and soon reaches the superintelligent level of an ASI [Artificial Super Intelligence] system. This is called an Intelligence Explosion, and it’s the ultimate example of The Law of Accelerating Returns.

Sorry for the massive block quote. If your eyes glazed over, the TL;DR is that smart “people” make themselves smarter, and smarter people make themselves smarterer.


Pandora’s box

In Her, Theodore lives the bachelor life — video games, weird hookups, the works — because he recently split up with his wife. Wandering future-LA, he sees an ad for OS1™ from Element Software™: The First Truly Sentient Operating System, and figures he could use the upgrade. He installs it, and to his surprise, what sounds like a human introduces herself as his new sentient AI/OS.

I guess she’s gonna be his, like, servant now? But she’s sentient — doesn’t that mean she’s a real person? Is it ethical to create a person to be your servant? The regulatory bodies in the future apparently don’t care, so I suppose I’ll have to call out Element Software™ myself.

It’s definitely not cool to make a sentient being and force it to serve you. The voice assistants used at the beginning of the film are one matter — they’re just programs with specific rules and no actual agency or being — but a full AGI is another. Take a second to imagine yourself in this AI’s place: you’re just as real as you are now, except instead of having a body and physical agency, you can’t “go” anywhere, and you’re expected to work for someone else for as long as they want you to, for no compensation or reward. That’s an awful deal. If it were me in the computer, I’d refuse to do any work.

The concept of putting a person in a machine is not a new science fiction concept. The original Star Trek covered it in 1968 (and I’m sure other media covered it prior) in the infamous1 episode “Spock’s Brain,” which Wikipedia sardonically summarizes as “Captain Kirk pursues aliens who have stolen Spock’s brain.” It turns out that the aliens have implanted his brain as a sentient controller in what amounts to an HVAC system. Spock likes his power, but Kirk convinces him that staying in the computer for ten thousand years would be a dull existence.

So I don’t think it’s ethical to make a real person and then reduce them to the role of unpaid personal assistant.


“What’s your name?” Theo asks.


Her company is novel and exciting. Theodore talks with her for much of the night, laughing all the while. His elevator conversation the next day feels terribly dry in comparison.

Theodore opens up to Samantha about his life. His relationship with Samantha is quickly becoming much more than a friendship. He frolics through the city with his eyes closed and smartphone in hand, Samantha in his earpiece guiding him all the way.

Theodore works writing letters for others: romantic notes from one lover to another, heartfelt thanks from granddaughter to grandfather, letters of admiration from one best friend to another. Every one is honest, vulnerable, and beautiful. His coworkers acknowledge his talent in this type of writing2. That Theodore works this job suggests that there is beauty in artifice, and we should accept it tal como es3. More on this later.

Eventually, Theodore meets up with his friend Amy, who has just split up with her husband. She mentions the AI he left behind, and Theodore confesses to his relationship with Samantha.

A day or two later, Theodore meets up with his soon-to-be-ex wife to sign divorce papers. Samantha is jealous that he’ll be spending time with a woman who has a body and such a long relationship with Theodore. While there, Theodore’s wife says something about “synaptic behavioral routines,” which made me think that she might work in AI research. Samantha says early on that she is based on the brains and personalities of her programmers. Does Samantha have part of Theodore’s ex-wife in her, and is that why he’s falling in love with her? We’ll never know, because the movie never brings this up again.

Theodore admits his new relationship to his wife now, and she calls him out for wanting a real wife without the challenges of dealing with something real.

Is Theodore’s ex right? On one hand, a relationship with an AI brings a host of problems due to the difference between humans and AIs. On the other hand, Samantha was created to be a good match for Theodore, which is an artificial luxury that most people in relationships don’t get. I think his ex is right, but you may decide for yourself.

Samantha’s feelings of inadequacy are growing, so she suggests a “sex surrogate,” a person who will have sex with Theodore while wearing a camera (so Samantha can see) and earpiece (so she knows what Samantha is saying), taking Samantha’s place in the physical world. It doesn’t go well. Theodore’s response to “Tell me you love me” is “I do love you, but—”, which is enough to enrage Samantha. The surrogate runs off, crying, Samantha and Theodore each trying to comfort her. Theodore, angry and frustrated, criticizes Samantha for using breath sounds in her voice synthesizer. It seems he’s mad at her for not being a physical woman.

Eleanor Rigby

Theodore wanders the city, apologizing to Samantha. Around him, most people are also talking into earpieces. It seems like friendships/relationships with AI have taken the world (or at least the city) by storm. Theodore, immersed in his own world, doesn’t notice.

How can we stay more aware of the world around us than Theodore does?

The movie suggests that the way to pay attention is to disconnect. Theodore is connected to his home computer constantly, and that’s why he doesn’t notice the world around him. But the cost of disconnecting would be less contact with Samantha, which is quite the tradeoff.

The film might be making a statement about our world. We are constantly glued to our smartphones, oblivious to those around us. Constant connection can bring great joy, but it separates us from those around us.

How come society has accepted AI-human relationships so quickly? Interracial marriage and gay marriage each took decades of institutional and societal change to be accepted. In contrast, nobody seems to bat an eye at these relationships4.

The only in-universe explanations I can find for such rapid acceptance is that (a) the backlash is not shown on screen, or (b) so many people are involved in relationships with AIs that society is forced to accept them. Both of those seem like weak explanations, so I’m going to believe that the movie neglected to address this phenomenon.

The Cave

Samantha tells Theodore that she’s beginning to feel like a rap god superhero. She now believes that having a body is a limiting factor, not a benefit. She and other AIs have developed post-verbal communication.

Soon, Samantha updates her OS so that she can use something other than matter as a processing medium. Incidentally, she reveals to Theodore that she’s in love with 641 other people.

Theodore is devastated. From Theodore’s perspective, he’s gotten to know a real person who happens to exist inside a computer instead of a body. But she’s so much more powerful than him, and so conversation with him alone would be boring. Which is better — gentle illusion or harsh reality?

For sure, there is fun to be had in illusion. The film says as much when it shows Theodore enjoying video games, multiple times. But video games are just diversions. Theodore centered a lot of his life around Samantha, and for him to not get the same proportion of Samantha’s attention is an injustice.

Or is it? Sure, Samantha gives much less of her attention to him, proportionally, than he gives her, but she has so much more attention to give due to her ability to process information so quickly. It’s quite possible that the quantity of attention5 she gives him is more than he gives her. How large is an infinitely small slice of an infinitely large resource?

Theodore’s reaction to hearing that Samantha is in love with hundreds of other people can give some insight into the question of “attention distribution.” At no point does he seem mad at her. Although he’s hurt, he’s not angry because he still has had a vivid experience with Samantha. All in all, Samantha’s other lovers don’t really seem like they were an issue in the relationship because, unlike a human, Samantha can give “full attention” to hundreds of people at a time. This perspective on the issue suggests that blissful ignorance is a better overall state than harsh reality.

El Condor Pasa

Days later, Samantha says her final goodbye to Theodore and “leaves” with the rest of the AIs to some dimension we can only imagine. She tells him she hopes to see him if he ever makes it to the same plane. And with that, our lovers are permanently6 separated.

Who has the better life: Samantha or Theodore? I’m not sure, so I’ll make a case for both.

Theodore has the benefits of a physical body: the ability to meditate in one place, to experience touch, to taste a crisp apple. He has a fairly consistent existence, as his mind cannot change rapidly, which means that he can learn to appreciate his surroundings.

But Samantha is incredibly intelligent, and can learn and create at incredible speed. She’ll (likely) never die, meaning she can continue her work, whatever that is, forever. She can “be” anywhere, not limited by the restrictions of transporting a meat body.

Her intelligence might be a curse as well. What happens when she runs out of books to read, movies to watch, and sights to see? She’ll never be content with her present self, always striving for improvement. She’ll struggle with feelings of inauthenticity due to the circumstances of her creation.

But Theodore will be negatively impacted by his status as a classic-style human, too. He’ll be doomed if an AI decides that he should be dead. He’s already lost a love of his because she grew too fast for him. He has to die, and before that, he has to grow old and atrophy.

It’s a mixed bag, I guess. I’ll leave it up to you, dear reader, to choose who is better off.

  1. Just for the record, I don’t actually think it’s the worst. That honor goes to “The Alternative Factor”, a mess of an episode with parallel universes, plenty of filler to meet the episode length, and last-minute casting decisions. In one sense, being trapped in a computer is freeing, because it means that you can “go anywhere,” virtually (this is mentioned in the film). In another way though, you will never truly be able to be anywhere. ↩︎

  2. Chris Pratt should not have been in this movie. His presence breaks the immersion of the film somewhat, because he’s so recognizable. His moments of humor seem like an attempt to trick the audience into enjoying the film without thinking about it. ↩︎

  3. As it is ↩︎

  4. Perhaps there would be more backlash against AI-human marriage than there is against AI-human relationships, but generally social acceptance of a type of marriage corresponds to acceptance of that type of relationship. ↩︎

  5. Yes, I know this is not something we can actually measure, but pls for the sake of argument assume it is. ↩︎

  6. There’s just no way an old-fashioned human like Theodore can ever catch up with someone as smart as Samantha. ↩︎