Technology major IBM is exploring the application of Generative Artificial Intelligence (GenAI) to new models beyond language and code, in a bid to create solutions for diverse market domains and areas, said Sriram Raghavan – Vice-President, IBM Research AI.
IBM has partnered with NASA and built an AI foundation model for geospatial data, which has various applications. It is also working with partners to apply models to understand chemistry that would have implications for therapeutics, medications and drug discovery.
“Implicitly, the models have been either language or code, but we are working to apply the same technology to other types of data as well. In all scenarios, we will supply the core technology, the partners and clients with domain expertise will bring them to the market,” Raghavan told businessline.
The company is also interested in applying Generative AI to increasingly more IT data including logs, events, and cybersecurity among others, which Raghavan explains have more structure than just text and foundation models built to leverage the unique characteristics of logs and events will be interesting. It is also exploring with clients to build models and sensor and time series data.
Raghavan explains, “The idea with foundation models is to take large amounts of unstructured data to create a rich representation. It so happens that right now language and code are in, which these representations seem to have immediate use cases and value. But in all of the other domains as well, the work is happening in the labs and I am confident that over the years, we’re going to see GenAI solutions pop up.”
Talking about the contribution of Indian teams, Raghavan noted that both software and research teams in India have played a big role in Watsonx — a new AI and data platform unveiled earlier this year. On the software side, the Automation Innovation Centre in Kochi is at the heart of a lot of work we do in Watsonx. Research labs are focused on work in conversational AI. Core technologies for Watsonx Assistant are built by Indian research teams, he noted.