Will artificial general intelligence really be the ‘last invention’?

I.J. Good proposed in the 1960s that the first ultraintelligent machine could trigger an intelligence explosion, suggesting it might be the last invention humans need to make.
The core assumption behind artificial general intelligence (AGI) becoming the last invention is questioned, particularly the idea that innovation is a smooth transition from idea to impact.
Discovery and invention are as much about navigating practical bottlenecks and translating ideas into real-world outcomes as they are about raw intelligence.
Even if AGI is superior to humans in cognitive tasks, it would still require human judgment and practical know-how to navigate complex, real-world situations.
The knowledge problem implies that not all relevant information is codifiable, and vital contextual, routine, and tacit knowledge is often embedded in human practices and difficult to replicate.
The claim that AGI will replace human roles entirely does not account for legal, ethical, and practical barriers to full automation and the transferability of human expertise.
Ultimately, as intelligence becomes cheaper, human skills in delivering valuable outcomes and managing practical constraints will become increasingly valuable. AGI alone cannot guarantee that practical and institutional infrastructures are aligned for widespread impact.

You may have missed