Official newspaper of The University of Texas at Austin

The Daily Texan

Official newspaper of The University of Texas at Austin

The Daily Texan

Official newspaper of The University of Texas at Austin

The Daily Texan

Advertise in our classifieds section
Your classified listing could be here!
October 4, 2022
LISTEN IN

Harvard professor suggests new approach to artificial intelligence

2017-03-30_Harvard_AI_Yifan
Yifan Lyu

While many people foresee computers replacing humans in some fields, Harvard professor Barbara Grosz said we should be working toward a more collaborative relationship with artificial intelligence at her Wednesday presentation.

Grosz said computers have changed immensely since their conception as a single system in one place with a single user. Now, multiple systems all over the world connect elaborate networks.

This change makes the idea of artificial intelligence different from what it used to be, Grosz said. The omniscient AI of science fiction is not the AI that she believes we should be working toward. 


“Humans have a centuries-long fascination and fear of artificial intelligence,” Grosz said. “We have technological fantasies and nightmares. I hope to bring some reality to people’s notions of artificial intelligence.”

In 1950, computer science pioneer Alan Turing developed the Turing test to see if a computer’s responses to a series of questions could make it indistinguishable from a human. Even modern computers are unable to pass it, Grosz said.

“We have enough humans already,” linguistics Ph.D student Chris Brown said. “We don’t need AI to be a replacement for them.”

Because of the technological development, Grosz said the goal of the Turing test should be altered. 

Grosz said she views the Turing test as too philosophical for what we should be aiming for. Her suggestion is that the goal for AI should not just be to get answers indistinguishable from a human’s. 

“What we should ask is, ‘Can a computer team member behave over the long term, in such a way that people on the team will not notice it’s not human or think it’s stupid?’” Grosz said.

Erik Lindgren, electrical engineering graduate student, said he believes professionals need to focus on implementing this approach.

“The next step is to apply (collaboration with artificial intelligence) to more areas,” Lindgren said.

Applications of the collaborative approach to AI include healthcare, education, rescue and rebuilding, writing and semi-autonomous driving, Grosz said. 

Grosz said AI systems can aid the complex network of professionals who work in health care and education by keeping the work organized and efficient, without needing the humans to consult each other. 

“It’s a very exciting time for AI,” Grosz said. “The most exciting time since I started in the field.”

More to Discover
Activate Search
Harvard professor suggests new approach to artificial intelligence