November 6, 2023 - 10:00am

On Sunday Elon Musk’s X AI team launched an early testing model of Grok, a new AI “for understanding the universe”. Grok will, according to the promotional blurb, “answer questions with a bit of wit” and “a rebellious streak”. It will also “answer spicy questions that are rejected by most other AI systems”, which we can assume means it will not have as many of the “safety” restrictions that currently forbid models such as Chat-GPT responding on taboo themes such as race.

Far more significantly, though, the development team boasts that Grok “has real-time knowledge of the world via the X platform”. That is: unlike other large language models trained on a set of text that — however large — is static and finite, Grok will base its knowledge of the world on the live, fast-moving hive mind that is the website formerly known as Twitter.

Will this work? Those who lament Twitter’s decline in the Musk era may be shaking their heads. Once beloved of elites and credentialled journalists who saw themselves as the elect moral guardians of the Right Side of History, a question mark has hung over Twitter’s status since the Musk takeover. Recent reports bemoan its decline, thanks to feature tinkering, algorithm tweaks, and bleed in ad revenue and respected users. They warn that the only possible direction from here is further down into exploitative “enshittification”, essentially milking a captive user base until the ratio of quality to garbage finally becomes intolerable and people leave.

Others argue that it’s simply too addictive to be easily wrecked, and will weather whatever Musk has planned. Now, Grok potentially introduces whole a new set of incentives and pressures. For Twitter really is a kind of collective intelligence, where you can watch consensus-formation occur in real time. More completely than any other digital platform I can think of, Twitter fits the description David Bowie gave of the internet in a 1999 Paxman interview: “an alien lifeform”. 

And if the point was always to use its collective consciousness to power an AI, its usefulness will only be as good as the hive mind from which its “knowledge” is drawn. This in turn provides — in theory at least — a strong incentive to hold back from changes to the platform that would aggressively degrade its operation as a collective intelligence, however much such changes might result in short-term profit. 

So plugging Twitter in as the backend for synthetic super-consciousness may be what saves it from being strip-mined to trashy extinction on the scrapheap of exhausted social media platforms. Who knows, this might even make up for the sci-fi creepiness of knowing that when we tweet we’re participating in the genesis of this entity. 

Perhaps more importantly yet, Musk’s latest move answers a question that’s been nagging me for a while: what will provide the divine spark for AI? So far no AI even remotely resembles an actual intelligence, only pattern-recognising machines that make suggestions based on existing datasets. They can be eerily effective sometimes, but are really no more sentient than the autocorrect on one’s phone. So I have long wondered: how does such a technology make the leap to actual awareness? 

Well, now we know. Providing Musk manages to resist the temptation to wreck the hive intelligence he just bought for $44bn by driving its most dedicated users elsewhere, it turns out that the answer may be: you.

Mary Harrington is a contributing editor at UnHerd.