LUDOVIC MARIN/Getty Images
© 2023 Blaze Media LLC. All rights reserved.
Elon Musk told British Prime Minister Rishi Sunak in an interview published on Nov. 2 that AI will eventually eliminate the need for human labor. Two days later, Musk's xAI Corporation announced the rollout of its "spicy" alternative to OpenAI's ChatGPT.
The company's new virtual assistant, apparently a precursor to some not-so-far-off master, represents a big play on Musk's part in a field he recently suggested has the "potential of civilization destruction."
While the South African billionaire has repeatedly raised the alarm about the dangers of artificial intelligence, he has also suggested that humanity's hope in surmounting these dangers lies in transcending the flesh. If a hopeful gamble, then "Grok-1" may ultimately be more than just another large language model with multi-modal capabilities and instead a step toward the "merger of biological intelligence and machine intelligence" teased by Musk in 2017.
Grok, which is heavily integrated with X, takes its name from sci-fi author and occultist Robert Heinlein's "Stranger in a Strange Land," where it was said to mean "understand[ing] so thoroughly that the observer becomes a part of the observed — to merge, blend, intermarry, lose identity in group experience. ... All that groks is God."
Despite this Terran definition from the 1960s, xAI's machine-learning neuro network is presently poised neither to eliminate the boundaries between the observer and their observations nor to divinize users. Rather, it will "answer questions with a bit of wit and ... a rebellious streak" and distinguish itself by answering "spicy questions that are rejected by most other AI systems" — although apparently it won't teach users how to make illicit drugs from scratch.
ChatGPT, by way of comparison, will shut down various requests leading to content its designers have deemed inappropriate. ChatGPT's answers are ostensibly also conditional on ideological sensitivities.
A study conducted at the University of East Anglia recently found that OpenAI's ChatGPT produced responses that favored Democrats in the U.S., the leftist Labour Party in the U.K., and socialists in Brazil.
"Our findings reinforce concerns that AI systems could replicate, or even amplify, existing challenges posed by the internet and social media," said lead author Fabio Motoki of the Norwich Business School.
While Grok may be open to unorthodox and provocative questions, it's not yet clear whether it too will amplify biases online in its answers. Given that one of its selling features is that it draws data from X in real time — tasteless memes, news headlines, and statist propaganda included — biases are sure to port over from those activists who failed to follow through on threats to flee to Mastodon as well as from everyone else. A dominant bias on X may similarly predominate on Grok.
Grok and symbiotes
Musk founded xAI years after his unsuccessful attempt to take over OpenAI. The company indicated in a blog post that Grok-1 is the product of only four months of development. The prototype was initially trained with 33 billion parameters. Since September, the developers claim to have made "significant improvements in reasoning and coding capabilities ... achieving 63.2% on the HumanEval coding task and 73% on [multidisciplinary multiple choice questions]."
One fiscal quarter in and Grok-1 is said to have already surpassed ChatGPT-3.5 and Inflection-1. The developers indicated that Grok is "only surpassed by models that were trained with a significantly larger amount of training data and compute resources like GPT-4."
To what end? Is Grok simply a means to eliminate the need for human labor, as Musk prophesied to Sunak? Or is it a means to a greater cataclysm than the end of work?
xAI Corp, whose compute Musk claims is "doubling every 2 to 3 months," is said to be working closely with X, Tesla, and other companies. As X, xAI, and Neuralink did not respond to requests for comment by deadline, it is presently unclear whether Neuralink is one of those companies. Regardless, whether now or later, such a powerful LLM used in conjunction with the Neuralink Application could conceivably help move the needle on Musk's goal of helping humanity "achieve symbiosis with machines."
To avoid becoming subordinate to the AI demigods of the near future, Musk claimed at the 2016 Code Conference that it might be prudent to develop "an AI layer. ... You've got your limbic system, your cortex, and then a digital layer, sort of a third layer above the cortex, that could work well and symbiotically with you."
Now, seven years later, he has a neurotechnology company with an implant that uses 1024 electrodes distributed across 64 tiny threads to record neural activity under the skull. Low-power chips and electronics process these signals, then send them to the external Neuralink Application, which decodes the data into actions and intents.
If Grok is the AI assistant it is cracked up to be, it might be able to help decode neural data and make for a more seamless brain-computer back-and-forth. Grok could end up becoming the third layer above the cortex.
Musk maintains that "in the long term, Neuralink hopes to play a role in AI risk [sic] civilizational risk reduction by improving human to AI (and human to human) bandwidth by several orders of magnitude."
The company is set to begin human trials of its implant. The forthcoming FDA-approved trials will assess the functionality of Neuralink's BCI for enabling paraplegics to control a computer cursor or keyboard using their thoughts alone. Perhaps Grok will one day make the cursor redundant and actualize the virtualized will without additional mediation.
Contrary to Heinlein's definition, rather than a collapse of observer and observed, Grok might come to mean a melding of enfleshed master and silicon servant.
It will be worth watching for what roles, if any, future iterations of Grok will have in bridging human thoughts and virtual actions — and whether the wisecracking AI's ultimate punch line will be processed directly by post-humans. In the meantime, xAI's new assistant will push the envelope of what's thought possible with an AI bot and criticize our breakfast preferences.
Want to leave a tip?
We answer to you. Help keep our content free of advertisers and big tech censorship by leaving a tip today.
Joseph MacKinnon is a staff writer for Blaze News.