Every five years, computers become 10 times cheaper – something that has been happening since 1941.
This is according to professor Jürgen Schmidhuber, director of the Artificial Intelligence Initiative at King Abdullah University of Science and Technology, and scientific director of Swiss AI Lab IDSIAI (Istituto Dalle Molle di Studi sull'Intelligenza Artificiale).
Schmidhuber this week delivered a keynote presentation on the future of self-improving artificial intelligence (AI) and neural networks at the second Global AI Summit in Riyadh, Saudi Arabia.
The professor’s AI lab’s deep learning neural networks based on ideas published in his paper, the "Annus Mirabilis" in 1990-1991, have revolutionised machine learning and AI.
According to Schmidhuber, who is known as the father of modern-day AI, technology has advanced so rapidly over the years that PCs today are significantly cheaper than they were in the past. The more they advance, the more likely they are to become even cheaper and thus, more accessible to communities that previously had no access to them.
Today, the processing speed of computers has multiplied, and as they get faster every year, they also become more affordable, he noted.
“When Konrad Zuse built the first program-controlled computer in Berlin in 1941, he could perform roughly one operation per second. But now, after many decades of getting a factor-of-100-per-decade, we can do almost a billion billions of operations per second – for a fraction of the price of a computer at the time. In 2012, a computer was 100 times more expensive than the price of an average computer in 2022.”
According to Schmidhuber, in the same way as cellphones are much cheaper today as they have evolved, there will be a revolution that will enable everyone, everywhere, to have access to AI, and almost all businesses will profit greatly from it.
“Very soon, we are going to have little computational devices, which can only compute as much as the human brain, but 50 years later that computational device will compute as much as all human brains on earth combined. And as you can imagine, everything is going to change,” he added.
Since age 15, Schmidhuber’s main goal has been to build a self-improving AI smarter than himself. Once this has been achieved, he hopes to retire.
Discussing the role of AI in improving human life, he emphasised the ability for an AI to improve itself over time, through a process called recursive self-improvement. This has enabled the technology to serve humanity, and allowed progress and qualitative leaps in many fields, especially in the medical and economic sectors.
Artificial networks, the development of machine intelligence, the empowerment of smart cities and the manufacture of robots have contributed to the service of humanity and the improvement of life, he continued.
In the medical fields, it has shortened the time and effort in diagnosing medical cases as accurately as possible, as well as in performing operations.
He explained that in the automotive industry, AI is now used on a larger scale than ever before, as autonomous tools can improve driving by determining lanes, speed and traffic congestion.
“The robots have now begun to self-learn after they have crossed many stages over the past decades. Today, we are witnessing the use of AI technology in the most difficult games − such as chess, which requires mental effort and high concentration.”
Share