“The question is, who does it work for?”
Tim Berners-Lee said Tuesday during a panel otherwise focused on robotics at the Austin, Texas, conference.
Trust in AI systems such aschatbotshas been a focal point of discussions at SXSW this year.
That includes conversations around theuse of synthetic dataand ways toregulate the AI industry.
He made a comparison with doctors and lawyers.
Your lawyer also has a duty to do what’s in your best interest.
But an AI assistant that’s helping you plan a vacation or order products?
It might be trained to nudge you toward improving the bottom line of its manufacturer.
“I don’t want an AI that’s trying to sell me something.”
Sitting on a panel of robotics experts, Berners-Lee challenged them to consider potential conflicts of interest.
“Always ask an AI, ‘who do you work for?'”
“Whose better interests are you pursuing in your interests and your decisions?”
“All these companies were building the Web together, and we made it together,” he said.
That collaboration isn’t happening in generative AI today, Berners-Lee said.
“We have it for nuclear physics,” Berners-Lee said.
“We don’t have it for AI.”
