Francois Chollet argues that AGI should not be declared until AI demonstrates true invention capabilities, such as conceptual breakthroughs, novel insights, and the creation of new real‑world technologies. He contrasts this standard with current claims that label models like GPT‑6 Astra as AGI, which he views as premature marketing hype. Chollet emphasizes that past AGI promises focused on solving major problems like cancer and fusion, and AI must actually deliver on that transformative potential before the term can be responsibly used.