When the top minds in artificial intelligence can't agree on whether general intelligence even exists, you know we're in for an interesting ride. On Monday, Tesla Inc. (TSLA) CEO Elon Musk jumped into a brewing dispute between two AI heavyweights, backing Alphabet Inc.'s (GOOGL) Google DeepMind chief Demis Hassabis after Meta Platforms, Inc.'s (META) outgoing AI leader Yann LeCun dismissed the entire concept of "general intelligence" as basically nonsense.

Musk Picks A Side

Musk didn't mince words, amplifying Hassabis' response to LeCun on X with a characteristically brief endorsement: "Demis is right." That's it. Two words that throw Musk's considerable weight behind one side of what's becoming a fundamental split in how the AI community thinks about intelligence itself.

The disagreement goes beyond typical tech industry squabbling. It cuts to the heart of whether artificial general intelligence—AGI, the theoretical AI that can match or exceed human capabilities across any cognitive task—is a meaningful goal or just marketing hype with a science fiction wrapper.

What They're Actually Arguing About

Hassabis came out swinging against LeCun's claim that "there is no such thing as general intelligence." According to Hassabis, LeCun is mixing up general intelligence with some impossible ideal of universal intelligence that can do literally everything.

In a detailed response, Hassabis pointed out that human brains are spectacularly complex systems designed to be general learning machines. Sure, nothing in the real world can escape some specialization—physics has rules, after all—but the human brain's architecture is fundamentally capable of learning any computable task, at least in theory, given enough time, memory and data.

His ace in the hole? The fact that humans evolved to hunt and gather, yet somehow managed to invent chess, quantum mechanics, and TikTok. That versatility, Hassabis argues, demonstrates genuine generality rather than narrow specialization.

LeCun's Counterargument

LeCun fired back, characterizing the whole thing as largely a semantic disagreement. His objection centers on equating "general" with "human-level," which he thinks misses the point entirely.

Humans, in LeCun's view, are actually extremely specialized systems that happen to be optimized for efficiency in specific domains. Yes, the human brain is theoretically Turing complete—meaning it can compute anything a computer can, given infinite time and resources—but it's spectacularly inefficient at most computational problems under real-world constraints.

LeCun also pushed back on the idea that brains are genuinely general, noting they can only represent a tiny fraction of all possible mathematical functions. True generality, by this definition, is practically impossible.

Why This Actually Matters

This isn't just academic hairsplitting. The debate exposes a major fault line in AI research over whether AGI is a realistic target or a fundamentally flawed concept.

The predictions are all over the map. Salesforce Inc. (CRM) CEO Marc Benioff, who's pouring billions into AI development, has said the reality of AGI falls well short of the hype surrounding it.

Meanwhile, Hassabis himself has previously estimated roughly a 50% chance that AGI emerges within the next five years. Anthropic co-founder Ben Mann has floated 2028 as a possible arrival date. OpenAI CEO Sam Altman has suggested AGI could show up during President Donald Trump's current term. Former Google CEO Eric Schmidt, speaking at an April fireside chat, called it reasonable to expect AGI somewhere between 2028 and 2030.

That's quite a range—from "it's overhyped" to "maybe next Tuesday."

The disagreement between Hassabis and LeCun highlights how even the people building these systems can't agree on what they're building toward. If the concept of general intelligence itself is disputed, how do we know when we've achieved it? Or whether we should even be trying?

For now, the debate continues, with Musk firmly in Hassabis' corner and the rest of the AI world watching to see whose vision of intelligence—general, specialized, or something else entirely—ends up shaping the technology that might reshape everything else.