AGI

Updated: 12/29/2023 by Computer Hope
HAL AI from 2001: A space odyssey

Short for Artificial General Intelligence, AGI is an artificial intelligence capable of understanding and learning at the same level as a human. Unlike all specialized AI (Artificial Intelligence) available today designed for a specific task, AGI would be able to do almost everything a human can do. As of 2024, true AGI has not been achieved, and some believe it can never be achieved.

Note

AGI may also be called strong AI or true AI. AGI was also featured as a top term in 2023.

If AGI were to be achieved, what could it do?

Theoretically, if AGI occurred, a computer would become indistinguishable from a human, as it could perform and learn new abilities like a human. However, because it's a computer, it would learn much faster and surpass the smartest human quickly. After gaining this intelligence, it could do almost anything a human could do, allowing machines to replace most human workers. Because of its learning capabilities, it's also likely that AGI could help solve most of the complex problems we face today. However, if AGI got into the wrong hands or it believed humans were unnecessary, things could be drastically different, as commonly portrayed in science fiction shows.

Why is AGI researched if it could harm humanity?

If put in the wrong hands, a lot of technology can harm humans. However, most technologies invented have done more good than harm. For example, while it's possible to do bad things on the Internet, most agree that the Internet has done more positive things for society. Although AGI would be the most revolutionary invention, which could lead to different outcomes, technology has reached a point where most believe not continuing AGI research would be worse. If everyone with good intentions stopped researching and developing AGI because of its potential dangers, it would allow only those with bad intentions to develop and understand AGI.

Is ChatGPT considered AGI?

No, although ChatGPT and other similar specialized AI developed using LLMs (large language models) give the appearance of AGI with some of their responses, it is not AGI.

AI, Artificial intelligence terms, Computer acronyms, Microsoft Copilot, TLA