Become a TechRadar Insider
One of the most important figures at the heart of the rise of AI is Jensen Huang, co-founder and CEO of Nvidia. His company manufactures the GPUs that excel in training and inference, and is positioning itself at the heart of the infrastructure and energy buildout. Like many prominent figures, Huang also has a take on the risks posed by this technology.
The existential risks of AI
There's no denying that, with any new technology, AI carries plenty of risks in its mass deployment and use. However, no other technology seems to have drawn the intensity in the warnings that AI has attracted. Speaking on the podcast No Priors, the Nvidia boss railed against the "doomer" narratives that have dominated.
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.
Famously, a collection of scientists and tech industry leaders signed an open letter in 2023 warning of the possibility that AI could lead to the extinction of humankind, and that it should be considered alongside other societal-scale risks such as nuclear war. Signatories included Elon Musk, Sam Altman, and Bill Gates. One notable absentee was Huang.
His perspective, which he outlined in the podcast, is that many of these fears are wildly overblown and far too influenced by the science fiction tales from the 20th century.
Are the risks really overblown?
Many of the fears associated with AI aren't tied to its capabilities today, but rather the more hypothetical artificial general intelligence (AGI). This form of AI will, in theory, replicate human intelligence, act autonomously, modify its own code, and set its own aims and objectives that it could work toward.
The possible outcomes are stark, but they remain unattained and likely won't be for at least 10 or 15 years, according to some, while others suggest it may happen much sooner. Either way, Huang is correct to suggest that current AI deployments are far too primitive and require too much human hand-holding for us to lose control. That said, agentic AI offers a new layer in the corporate attack surface, the more it is adopted by organizations, and may be vulnerable to exploitation in the future.
