In the 1987 Nintendo game “Zelda II: The Adventure of Link”, there is a village you can visit and a house you can enter and a man in a purple shirt you can talk to. But the only thing he says is “I’m a mistake.” I was surprisingly delighted by this dialogue when I first encountered it on a Gamecube emulator back in 2009 (I’m not that old). It seemed both meta and existential, or just a fun bug. It turned out that Error was actually the character’s name, a sort of mistranslation in a joke that has since become a meme akin to “Your princess is in another castle” or “It’s dangerous to go alone! Take that” from the original “Legend of Zelda.”
Error is a non-player character (NPC), that is, a character that the player cannot control but is usually more than just video game wallpaper. Nearly 40 years later, NPCs have come a long way, with increasingly complex behaviors and conversational dialogue that extends well beyond three words. The upcoming game “Grand Theft Auto VI” is rumored to feature unprecedentedly complex NPCs with rich, fleshed-out personalities and stories, as video game companies explore the use of artificial intelligence to allow players to train NPCs as they interact.
In many ways, NPCs represent how many people have been introduced to AI, albeit in a relatively simplistic form compared to the chatbots and massive data center pigs we either love or hate, with little room in between. Years ago, NPCs became a slur that emerged from 4Chan, the anonymous online forum, to describe real people who don’t think for themselves and follow a set of cultural agendas, usually exemplified by a gray, expressionless avatar. The meme was popularized by chronically online cohorts of Donald Trump supporters during his first presidential term, alongside the idea of ”Trump Derangement Syndrome,” both implying that people oppose the MAGA movement out of reflexive unconsciousness and not, you know, because they had looked under the hood and seen it for what it is.
This may seem like a new way to call someone an idiot, but the NPC label has much deeper and frightening implications. On the surface, this seems like harmless video game humor, and I’m sure we’ve all had encounters with real humans that felt like interacting with tautological machines programmed with limited responses. “The weather is nice today. Have you heard of fame? Sports.” And it is certain that social networks, now invaded by robots, increasingly give the impression that each conversation takes place between different viruses. These realities do not correspond to all experiences, but following NPC slanders to their logical conclusion can be dangerous: declaring that a person’s supposed lack of inner reflection is justification for rejecting their humanity. After all, they’re just stupid characters in a video game (and maybe reality is just The Matrix, whatever), so it’s okay to ignore them, disenfranchise them, or abandon them to a dark fate.
Start your day with the essential news from the Show.
Sign up for our free morning newsletter, Crash Course.
Solipsism, the philosophical position that you can only experience your own consciousness and therefore others might not exist or have a soul, is not a new idea. René Descartes believed that animals were soulless, mindless automatons, although modern neuroscience has repeatedly refuted this hypothesis. (In a way, “I am a mistake” echoes Descartes’ famous quip about thinking and therefore acting.) But now that complex AI has spread from video games to the general public, it’s particularly dangerous to see the NPC label entrenched in mainstream politics, whether explicitly or not.
All of this has its roots in Donald Trump’s recent campaign to rebrand artificial intelligence as “super intelligence” via executive order, posting on social media on Thursday that “The White House considers anyone who uses the term ‘Artificial Intelligence,’ as opposed to the new, highly accepted and more accurate term, ‘Super Intelligence,’ to be the ENEMY! The idea was quickly embraced by technology leaders, with SpaceX CEO Elon Musk declaring this week that AI was “no more” and promising to change the name of his AI company, SpaceXAI, a SpaceX subsidiary, to SpaceXSI. Most recently, Amazon founder Jeff Bezos shared his approval in an interview with Fox News on Wednesday. “I like SI, because ‘artificial’ is not a very flattering thing. Like, no one wants artificial sweetener, artificial flavor,” Bezos said. “Artificial somehow means fake.”
Yes, most of us will roll our eyes at AI being “great,” but if these attempts at standardization gain traction and are pushed to their absurd conclusion, they could end up giving AI more rights than humans. (It’s worth noting that Musk often invokes the NPC meme.) It’s right there in the name: “superintelligence” is better than the ancient human form, so why shouldn’t we give rights to these computer programs and defend them from regulation? This is why stories about AI robots being placed in virtual torture chambers are not only idiotic, but promote the false idea that AI is capable of suffering. (Maybe you don’t put robots in rooms like that anyway – at the very least, it’s a waste of energy.)
After all, AI will either become the savior of humanity or it will make humans obsolete, depending on the day of the week. Of course, this technology is nifty, even if it causes enormous environmental destruction. But we are still far from elevating technology above ordinary human intelligence. But as with two-for-one, if you can be convinced of the myth that AI is superhuman, it goes along with the idea that ordinary humans are disposable. In other words, the tech bros are trying to turn us all into NPCs, into errors, notably through mass surveillance programs used to track migrants and spy on protesters.
Yes, thankfully, granting greater than human rights to so-called sentient AI is still a fringe idea, but we can see this changing before our eyes, not to mention we’ve already kind of done it with corporations. Recently, OpenAI CEO Sam Altman told POLITICO that “the world should accept that bad things happen for the sake of the benefits of this technology and the people who have agency.” To further consolidate its power – or at least prevent the economy from collapsing – Silicon Valley wants us to believe that its latest technological invention is capable of thinking, reasoning and moralizing. It’s no surprise that tech pundits have recently been exposed for trying to convince none other than the Pope that AI is conscious, but to his credit Leo
By spending time with large language models like ChatGPT and knowing a little about how they work, it becomes clear that it is just a very good computer program capable of reflecting human expression, sometimes terribly and sometimes surprisingly well. There remains no concrete evidence that AI algorithms are sentient.
Yet a growing number of people are convinced otherwise, as the controversial phenomenon of AI psychosis shows, which makes me wonder how so many people could have convinced themselves that an inflated version of “I’m a mistake” is a conscious being. It may feel like human interaction typing back and forth with these things, but there are always gaps where you can sense the uncanny valley, where you know it’s just a lot of math and nothing more. Maybe if you want to be really cynical about it, you could also boil down all of human nature, biology, and the emergence of consciousness to “a lot of math.” But our brains are not computers and physics has not yet fully explained consciousness – which is why they call it a hard problem. Instead of delving into this mystery, technicians inverted the metaphor, attributing personhood to robots while denying it to living humans. Talk about soulless.
Learn more
on artificial intelligence
Gn usa
