Mani Revists AI Grand Challenges in New Article
Wednesday, April 14, 2021 - by Aaron Aupperlee
Chess turned out to be an easy one. Translating speech in near real-time is mostly done. The accident-avoiding car? Maybe halfway there.
In 1988, Raj Reddy, the Moza Bint Nasser University Professor of Computer Science and Robotics at Carnegie Mellon University, proposed an ambitious set of grand challenges for artificial intelligence to tackle in the coming 30 years. With that deadline past, Ganesh Mani, Reddy's colleague and an adjunct faculty member in the Institute for Software Research, took a look at the status of the challenges and proposed a new set of tasks in a recent AI Magazine article.
"Grand challenges are important, as they act as compasses for researchers and practitioners alike — especially young professionals — who are pondering worthwhile problems to work on, testing the boundaries of what is possible," Mani wrote in "Artificial Intelligence's Grand Challenges: Past, Present and Future."
In the article, Mani laid out his challenges for AI in health, wealth and wisdom, proposing a shorter timeline than 30 years to match the upped speed of innovation. The new challenges are specific enough for researchers to conceptualize solutions and broad enough to enable use in a variety of other tasks.
Throughout his article, Mani scatters essays from top technologists on future applications of AI and other new challenges. Among them are Frank Chen, partner at Andreessen Horowitz; Steve Cross, retired Georgia Institute of Technology faculty member and former director and CEO in the Software Engineering Institute; Vanathi Gopalakrishnan, associate professor and director of the Intelligent Systems Program at the University of Pittsburgh; Ken Stanley, research manager at OpenAI; and Thomas Kalil, chief innovation officer at Schmidt Futures. Francesca Rossi, president of the Association for the Advancement of Artificial Intelligence and an IBM fellow, proposed in her essay an ethical challenge for AI that, in part, would require an AI system to evaluate not only its own behavior but the behavior of humans working with it.
"Grand challenges can be very inspirational for researchers and practitioners. Often the path to the result is more important than the result itself. Even before the challenge is achieved, many new technical, methodologies, and general lessons can be derived; and these can be reused or adapted in other contexts, leading to advancements toward other challenges as well," Rossi wrote.
Mani said the idea for the article started a couple of years ago during a chat in Reddy's office about the status of the grand challenges he laid out in 1988. Developing an AI to beat a world chess champion turned out to be relatively easy, the two of them agreed. And while Mani felt that translation technology from Google, Facebook and Microsoft had accomplished the challenge of developing a translating telephone allowing people speaking different languages to talk seamlessly, Reddy demurred, pointing out that the technology is not widely accessible, for example, to a rural farmer in an emerging economy without an internet-connected smartphone.
"Raj started this thread 30 years ago," Mani said. "This article was a way of continuing that conversation and advancing the technology to work with humans for the betterment of humans."
For more information, Contact:
Aaron Aupperlee | 412-268-9068 | aaupperlee@cmu.edu