India, Sept. 29 -- As artificial intelligence becomes embedded in publishing and knowledge workflows, access to powerful models is becoming less of a differentiator. The bigger challenge is ensuring that AI systems can work with reliable information, understand editorial rules and produce outputs that people can verify and trust.

For publishers, this challenge is particularly significant. AI can automate manuscript screening, language checks, classification and other repetitive processes, but decisions around research integrity, authorship, relevance and evidence still require human judgement. The underlying debate is no longer simply what AI can automate, but where automation should stop.

In an interview with Dataquest, Sriram Subraman...