What is “artificial general intelligence,” the cornerstone that obsesses big tech companies

Nvidia CEO Jensen Huang surprised everyone by declaring that Artificial General Intelligence (AGI) could arrive in 5 years and that it will be “8% better” than human intelligence.

IAG or “strong AI”, which for many researchers is a myth, is a system in which machines learn to make their own decisions. logical decisions. The theory indicates that, They will be able to surpass the human being in all aspects of knowledge.

These machines could develop their own feelings and have a subjective experience possess, a capacity that opens an ethical debate and which has been widely explored by cinema and science fiction literature.

The great fear is that AGI improves itself at an exponential rate until it reaches a point where its intelligence operates at a level unattainable by human understanding.

At Nvidia’s GTC conference, Huang set a deadline of just five years for companies to approach this possibility. He also took the opportunity to clarify some expectations generated by this technological change.

For the manager, predicting when general AI will appear depends largely on what should be understood as acceptable AGI. That’s why he says you need to be specific about what an AGI means and what you expect it to do to evaluate whether it has achieved its goal.

When it comes to tests, a general AI could successfully pass law exams, logic tests, or ability to solve college entrance exams.

“If you think of AGI as something very specific, a series of tests where a software program can perform very well (or maybe 8% better than most people), we will get there within five years,” he concludes Huang.

One of the problems with current models is how to resolve AI hallucinations and the tendency of some chatbots to do so invent answers they seem credible, but they are not based on facts.

For Huang, the solution is simpler than it seems. The solution is to make sure the answers are well researched. An “augmented catch-up generation” would solve the problem, as the approach is similar to that of basic literacy: examining the source and context.

Nvidia is a leading company in graphics processing technologywhose GPUs are crucial in the development of artificial intelligence because their cards allow a efficient processing of large amounts of data in artificial intelligence applications.

One of the first to talk about AGI was Marco Zuckerbergwho seems quite confident that his company will reach (sooner or later) this unprecedented technological milestone.

IAG: doubts and uncertainties

Not all AI experts are convinced of the arrival of general artificial intelligence as it is presented in science fiction.

The fluidity with which a chatbot operates or its ability to solve tests through data confirmation and the use of language is not proof that it is a being with a mental structure similar to that of a human.

For this group of researchers, a scenario in which general AI is realized is not possible because, in principle, there is no consensus on what is meant by that phase of AI. This is a philosophical question and not a scientific one, they say.