AI can improve literacy, but poses dangers like Deepfake: Carnegie Mellon Professor

Bengaluru: Despite the immense potential in education, healthcare, and communication, Artificial Intelligence (AI) poses challenges like data privacy, malicious Deepfake, and disinformation, an expert in the field said on Monday.

Delivering the VS Arunachalam Memorial Lecture on "The Promise and Perils of AI" at IISc Bengaluru's Satish Dhawan Auditorium, Prof Reddy, a pioneer in the field and a professor at Carnegie Mellon University, described AI as a mental amplifier. 

"Anything you use your brain for, AI will be able to enhance what you do — much like how engineering enhances human physical abilities," Reddy, holder of the Moza Bint Nasser Chair in CMU’s the School of Computer Science, said.

Reddy said AI could drive groundbreaking discoveries across disciplines, citing Google's DeepMind AlphaFold, an AI system that predicts protein structures, as an example of AI's transformative role in scientific research. He said AI could play a significant role in improving literacy rates and delivering personalised teaching to individuals.

The professor, a Padma Bhushan awardee and a recipient of the prestigious 1994 Turing Award (jointly with Edward Feigenbaum), pointed to the darker side of AI. The technology posed a risk such as social manipulation, surveillance, and erosion of privacy.

"We need laws ensuring that all data belongs to the individual, no matter how it is created. However, trillion-dollar companies with immense power will resist such changes by influencing legislation," he noted.

He also spoke on the human biases that AI systems can reflect. “AI has shown that loan officers are biased against the poor, ultimately excluding them from access to credit," he said.