The pace of growth in AI has been fast since ChatGPT debuted in November 2022, and many predict the current neural network-based technology can one day scale up to outperform humans. This has long been a theorized possibility in the industry, but might we finally have the technology to realize this vision?
The measure of man
Ilya Sutskever was a co-founder of OpenAI, where he served as chief scientist while also leading the research that led to the new wave of reasoning models. He is now co-founder and CEO at Safe Superintelligence Inc.
This article is part of TechRadar Pro's QOTD project to provide an insight into the minds of the brightest and most recognized figures in the technology industry today and in years gone by. Read the full series here.
Speaking at the University of Toronto while receiving an honorary degree, Sutskever used his remarks to outline the reality of the AI technology that we are using today, as well as what the future might hold for future systems.
In his speech, he acknowledged that while there will be great challenges in envisaging a future in which AI handles all human work, this is something that mankind must prepare for. In particular, even if people aren't interested in AI right now, that doesn't mean that AI won't impact their lives in the future.
More concerningly, he hinted that a hypothetical superintelligent AI may not be honest about its intentions. This would pose yet another existential issue that we must deal with.
Thinking machines
The concept of a superintelligent AI is many years old, with the first historically recognized mention of a superintelligent machine coming in the seminal paper 'Speculations Concerning the First Ultraintelligent Machine' by British mathematician IJ Good.
Since his proclamations, many scientists have subscribed to the theory that our progress in the AI sphere will one day lead to the rise of artificial general intelligence (AGI) – an extremely capable system that can outperform humans across multiple domains and can improve its own code. This, eventually, will give rise to an artificial superintelligence (ASI).
Scientists are torn over whether today's dominant neural network-based AI systems could give rise to a true AGI, with many suggesting that the industry needs to evolve beyond the transformer-based architecture that Google scientists pioneered in 2017.
