In the fiercely competitive landscape of artificial intelligence, IBM has introduced new generative AI models and features to enhance its recently-launched Watsonx data science platform, as it strives to maintain its relevance.
The fresh models, known as the Granite series models, appear to be substantial language models (LLMs) akin to OpenAI’s GPT-4 and ChatGPT, capable of summarizing, analyzing, and generating text. IBM has divulged minimal details about Granite, making comparisons with other LLMs, including their own, challenging. Nevertheless, the company has committed to disclosing the training data and processes leading up to the models’ availability in Q3 2022.
In the interim, Tarun Chopra, IBM’s VP of product management for data and AI, offered some insights via email:
“These new IBM Granite series models have been developed using curated, high-quality enterprise data rather than publicly scraped data,” Chopra explained. “The series encompasses specialized subsets within different domains, such as finance. This allows AI developers to use more efficient, domain-specific models for tasks like summarization, content generation, and insight extraction, as compared to larger general models.”
Within Watsonx.ai, a component enabling customers to test, deploy, and monitor models post-deployment, IBM is launching Tuning Studio. This tool empowers users to customize generative AI models to their specific data, requiring as few as 100 to 1,000 examples. Once users specify a task and provide labeled examples in the required format, they can deploy the model via an API from the IBM Cloud.
Additionally, Watsonx.ai will soon introduce a synthetic data generator for tabular data, commonly found in relational databases. IBM claims that generating synthetic data from custom schemas and internal datasets can aid in extracting insights for AI model training and fine-tuning with “reduced risk,” although the exact meaning of this phrase remains unclear.
Watsonx.data, IBM’s data store, will receive new generative AI capabilities starting in Q4 2023. These capabilities will enable users to streamline interactions with their data, potentially offering an experience similar to ChatGPT but focused on data visualization and transformation.
Dinesh Nirmal, IBM SVP of products, emphasized the company’s commitment to supporting clients through the entire AI lifecycle, from foundational data strategies to model tuning and governance.
IBM faces mounting pressure to demonstrate its competitive edge in the AI sector. In its second fiscal quarter, the company reported lower-than-expected revenue, prompting its CEO, Arvind Krishna, to stress the importance of AI and hybrid cloud technology, including Watsonx, in IBM’s growth strategy. IBM has secured over 150 corporate customers for Watsonx since its July rollout, including notable names like Samsung and Citi.
