Super-sized AI: Is a truly intelligent machine a matter of scale?

With a huge neural network that writes incredibly fluently, some experts suggest that scaling up current technology leads to human-level linguistic competence and ultimately true machine intelligence.


October 6, 2021


Pablo Hartado de Mendoza

when Artificial Intelligence GPT-3 was released last year, The impression of mastering humans was good language, Command to generate a fluent text stream. As the world progressed, veteran observers pointed out many of its mistakes and simple architectures.However
They claimed it was a clumsy machine. Except for reasons to believe that AI like GPT-3 may soon develop human-level language skills, reasoning, and other features of what we think. Intelligence..

The success of GPT-3 has come down to one reason. It was bigger than any AI of that type. So, roughly speaking, it boasted more artificial neurons. No one expected this change in scale to make such a difference. But as AI grows, it not only proves human unity in all kinds of tasks, but also demonstrates the ability to tackle challenges never seen before.

As a result, some people in the field are beginning to believe that relentless impetus for larger scales will lead to AI with human-like capabilities. Samuel Bowman of New York University is in it.. “A significant scale-up of current methods will make it easier to achieve human-level language behavior, especially after 10 or 2 years of computing improvements,” he says.

If true, it would be huge. Few experts thought that mechanical intelligence would arrive as a mere exercise in engineering. Of course, many still suspect that this is the case. You’ll know over time. Meanwhile, Bowman and others have superscale AI … Super-sized AI: Is a truly intelligent machine a matter of scale?

Back to top button