Thomson Reuters is developing its own generative AI foundation model described as general, intelligent and broadly capable but specifically enhanced for professional work.
While organizations for years now have been experimenting with AI, only a minority have seen true return on investment, said Jonathan Richard Schwarz, Thomson Reuters' head of AI research, during a preview webinar. This is why Thomson Reuters' generative model is being filled with the kind of highly specialized knowledge that general public models simply lack in order to prime it for professional tasks.
"This is really, in my eyes, a beautiful collaboration between the scientific excellence and the frontier thinking of academia and the scientific world, with the deep domain expertise, the incredible data, the subject matter feedback from people that know their domains in the best way," he said.
He noted that large language models generally have a vast breadth of training data, which often clusters in certain areas depending on the nature of the nature of the AI. Thomson Reuters is layering its own research and knowledge on top of the general data set in order to encourage both flexibility and productivity, as well as additional expertise from outside academic specialists as part of its Frontier AI Academic Lab, which is composed of more than 30 PhD-level researchers plus six tenured professors.
"This isn't your low-quality forum for recipes for cooking pies or whatever. This is the content that the firm has spent decades preparing," said Schwarz.
What this generally all translates into is that the model becomes capable of solving hard reasoning problems while maintaining accuracy. Schwarz noted that the work is not yet complete, promising there is much to come in terms of integration into other products and modes.
"We're making this a broadly capable model across professional domains," he said. "We're looking at full integration into the agenda pipelines. The current system is already probably on par with OpenAI overall, but significantly better factuality. There's increasing product integration into a wide range of products, and we're hoping to launch the model mid year, officially."
