IBM’s Mellea Tackles Open Source AI’s Hidden Weakness

RALEIGH, N.C. — Nathan Fulton remembers the moment he realized open source AI had an infrastructure problem.

Senior engineers kept asking him the same question: When would IBM‘s AI models support structured outputs? The feature, which enables developers to control how language models format their responses, had just been added to OpenAI‘s API, and suddenly everyone wanted it.

Fulton, a researcher and engineering manager at IBM Research’s MIT-IBM AI Lab in Cambridge, Mass., was baffled, he told The New Stack. The capability had existed…