Former Google CEO Eric Schmidt is worried about artificial intelligence. Here’s why

On being asked to explain the perils and promise of AI, Schmidt replied “Well imagine a world where you have an AI doctor that makes everyone healthier in the whole world. Imagine a world where you have an AI tutor that increases the educational capabilities of everyone in every language. These are remarkable and these technologies which are known as Large Language Models are clearly going to do this.”

Also Read: Why Generative AI is a house divided

However, the former Google CEO was quick to point out the threats that humanity faces from these language models. 

 “We face extraordinary new challenges from these things, whether it’s deep fakes… or people falling in love with their AI tutor,” he added. 

Elaborating on the things that make him worried Schmidt added that he is worried about the use of LLMs in biology, cyber-attacks and manipulating the way politics works. 

Schmidt also pointed out the speed at which these new artificial intelligence technologies are changing the world, noting that it took Gmail five years to reach 100 million daily active users, while ChatGPT reached the same milestone in about 2 months.

This is not the first time that Schmidt has raised the possibility. During an earlier interaction with author and journalist Walter Isaacson he had noted that large language models could be used for biological warfare and change the dynamics of war. 

He also raised the possibility of some people with access to raw models using them for potentially dangerous activities. 

“In a situation where all of the software has been released, these are what are called Raw models, which are unconstrained and the people who’ve played with the raw models say that these are ones that you and I can’t get to as normal users and they’re very frightening,” Schmidt said. 

Also Read: AI experts worried about Musk-led call for research pause.

According to Schmidt, someone with access to raw models could potentially ask the large language models to build a copy of the 1918 bird flu virus or find a way to blow up a building and hide the bomb.

The former Google CEO also said that we don’t have a philosophical basis for interacting with an intelligence that’s near our capabilities but non-human.