The article proposes a definition of AGI focused on advanced intellectual abilities for scientific discovery and asks for predictions on when it will be achieved.
A lot of people seem to have completely different definitions of AGI, so I want to clarify what I mean. For me, AGI would be an AI with the intellectual abilities needed to do advanced human cognitive work, but at a level beyond humans. It should be able to reason, analyze, learn, solve completely new problems, generate and test hypotheses, plan, run experiments, and independently conduct scientific research and make genuinely new discoveries. For example, imagine a scientist trying to discover a new treatment for a disease. The process might look roughly like this: 1. Gather and understand existing knowledge and data. 2. Notice an unexplained pattern, problem, or contradiction. 3. Generate possible explanations or hypotheses. 4. Work out what each hypothesis predicts. 5. Design experiments or simulations to test them. 6. Analyze the results and compare them with the predictions. 7. Reject, modify, or improve the hypotheses. 8. Repeat the process until a new explanation or solution is discovered. That is the kind of intellectual process I mean when I talk about AGI. Think of it like an insanely intelligent scientist who has no emotions or subjective experience. It doesn’t need to feel anything. It just needs to be extremely good at thinking and discovering. That’s what I personally think we actually want from AGI. We don’t really need an AI that can tell us jokes or have human emotions. We want something that can help us make breakthroughs in medicine, physics, mathematics, engineering, biology, and other fields, potentially going far beyond what any individual human scientist can do. So with this definition, what year do you think we’ll reach AGI? And what technologies or breakthroughs do you think will be necessary to get there by your predicted year?
The article argues that AGI has already been achieved based on historical definitions, as LLMs outperform most humans in intelligence, and suggests that ASI is the next frontier.
The article questions the meaning of AGI as claimed by AI labs, noting how the term has evolved and including expert opinions on its definition and implications.
The author reflects on the dramatic acceleration of AI capabilities from 2024 to 2026, highlighting breakthroughs in coding, mathematical reasoning, and a notable exploit by GPT 5.6 Sol, and predicts AGI may be closer than expected.