Q & A
1 minute read

How close are we to real AI reasoning?

IBM Research Fellow and AAAI President Francesca Rossi discussed the changes she’s seen in the field during her term, including the rising prominence of AI reasoning models.

Francesca Rossi looks at the camera with her arms crossed. Over her shoulder, text reads "Can AI models actually reason?"

The field of AI has developed rapidly in the two years since Francesca Rossi became the president of the Association for the Advancement of Artificial Intelligence (AAAI). Hardware and computing resources have become much more prominent, for example, as have evaluation benchmarks. As Rossi, an IBM Fellow, approached the end of her term as AAAI president, she met with Distinguished Research Scientist Murray Campbell to discuss the changes she’s seen in the field.

One of the biggest developments, Rossi told Campbell, is the rise of so-called reasoning models — including what it means for AI to reason, whether AI truly can reason like humans do, and why these questions matter. From the earliest days of AI, reasoning was programmed in. But over the decades these static lines of code have given way to emergent capabilities, to the degree that we now must ask how much autonomy to give AI systems.

How much can AI perform introspection? How confident can AI be in its own answers? Can AI become more aware of ethics? Watch the entirety of Rossi and Campbell’s discussion, where they discuss new reasoning capabilities and what they mean for the future of AI research.

Date