In a recent interview, Aidan Gomez, CEO and co-founder of Cohere, stated fears of an AI takeover are unfounded due to its reliance on human training data and the limits of exponential growth.
He explained, “I think I’m empathetic to the fears, you know, the sci-fi narrative of computers or AI taking over and destroying the world. It’s been going on for decades, and so it’s really deeply embedded within our culture. It gets lots of clicks, headlines, it gets attention. It’s shocking. I understand why people are scared of it and why some say it to get attention.”
Gomez highlighted that this is not a technical truth of the technology and that continuous exponential scaling does not happen. There are friction points and complexities. He also mentioned that the intelligence of these models is limited by the humans who create them, as it is our data and knowledge that teach them.
However, he believes real risks lie in deploying AI in high-stakes scenarios like medicine and advocates for scrutiny and tough discussions on AI deployment, rather than sensational sci-fi narratives.
What is Cohere Upto
In the interview, he also discussed how the original goal of the project was to improve Google Translate, a very well-known problem. He noted that it has been extraordinary to see the broad impact of a technology developed to enhance translation. Gomez was also the co-author of the original Transformers paper which forms the crux of today’s generative AI products.
Recently, AIM spoke to Saurabh Baji, SVP of Engineering at Cohere, about the mixed emotions in Silicon Valley regarding achieving AGI, as seen in the recent banter between Meta’s Yann LeCun and xAI’s Elon Musk.
“We remain concentrated on designing AI solutions that deliver better workforce and customer experiences for businesses today rather than pursuing abstract concepts like AGI,” said Baji.