AGI is meaningless, it also already exists

AI, AGI, or ASI are essentially umbrella terms for "everything computers can’t do yet."

But if we truly dig into definitions, today’s models are already generally intelligent in significant ways: large language models (LLMs) routinely answer most general questions across countless domains. They might not always provide innovative or optimal solutions, but they reliably offer useful, contextually relevant responses.

Even before LLMs, we’ve seen forms of broad general intelligence emerge. Consider Google or the internet itself—systems broadly intelligent enough to help users find answers, navigate complex queries, translate languages, provide accurate maps and directions, and much more. General intelligence doesn’t necessarily require a single executable; distributed systems have exhibited this trait for decades.

Regarding "super intelligence," we already have narrow superintelligence. Calculators have long surpassed human capability in mathematical speed and accuracy. Chess computers surpassed humans in strategy decades ago. Search engines and recommendation systems already demonstrate superhuman abilities within narrow tasks.

What we might not yet explicitly have is general superintelligence. Yet, in reality, combining existing generally intelligent systems (like LLMs) with narrowly superintelligent tools (calculators, search engines, maps, etc.) already approximates general superintelligence. Perhaps what’s missing is merely the perfect user interface that seamlessly integrates these capabilities into one cohesive system.

A human sitting in front of a computer or interacting with an LLM, leveraging search engines, calculators, and various tools, effectively embodies a form of artificial general superintelligence. Humans today are significantly augmented by technology, capable of performing tasks and solving problems they couldn’t previously manage unaided.

In conclusion, terms like AI, AGI, and ASI are largely meaningless or, at best, marketing terms useful for fundraising and hype. Just as "AI" finally became accepted as something that already exists, terms like AGI and ASI will follow suit. Human beings today already exist in a world enriched with artificial general and superintelligence.

This doesn’t mean these systems won’t continue to improve dramatically—they will—but we already have “AGSI”.

Whatever that means.

Originally published on X