Video produced by Steve Nathans-Kelly
In this clip from Data Summit 2026, Cal Al-Dhubaib, principal engineer at Rubrik, explores how data complexity and task complexity increase uncertainty when working with AI systems and trusting the quality of the summaries they provide, the actionable code they generate, the biases they reproduce, and the resulting confusion, and how these challenges multiply in the transition from generative to agentic AI.
“What we’re getting with these AI systems is an explosion of edge cases,” Al Dubaib says. “Machine learning and AI systems are known to be vulnerable to very simple manipulations.”
He explained that these models are highly sensitive to biases and errors in the real data. Context can also be compromised or manipulated.
“In large-scale contexts, the language model itself cannot disambiguate from contradictory sources,” Al-Dubaib said.
The Annual Data Summit Conference was held in Boston on May 6-7, 2026, with a pre-conference workshop held on May 5th.
Videos and clips of Data Summit 2026 presentations are now available on-demand. DBTA YouTube Channel.
More information about Data Summit 2027 coming soon.
