IBM CEO Champions True Open Source in Enterprise Generative AI: New Initiatives Unveiled at Think 2024

IBM is enhancing its generative AI initiatives with new technologies and partnerships announced at the Think 2024 conference.

With a legacy in AI that spans decades, IBM previously introduced the Watsonx generative AI platform at Think 2023. This platform serves as the backbone of IBM’s AI endeavors, offering organizations enterprise-grade models, governance, and tools. At Think 2024, IBM has unveiled its Granite models, now fully accessible as open-source code. These models, ranging from 3 to 34 billion parameters, support both code and language tasks. Additionally, IBM will integrate Mistral AI models, including Mistral Large, into its platform.

Among the popular applications for generative AI is the development of AI assistants, a focus for IBM as it rolls out new Watsonx assistants.

"AI presents a massive opportunity," said IBM CEO Arvind Krishna during a press roundtable. "Its impact will be comparable to that of the steam engine or the internet."

The Importance of Open Source in IBM's Granite Enterprise AI

IBM first introduced its Granite models in September 2023, continuously expanding its offerings. A notable model is a 20 billion parameter base code model that powers IBM’s Watsonx code assistant for Z service, aiding organizations in modernizing outdated COBOL applications.

What sets this announcement apart is the introduction of advanced Granite models under the open-source Apache license. While many vendors claim to offer open models, few comply with an Open Source Initiative (OSI) approved license. IBM asserts that only Mistral and Granite represent high-performance large language model (LLM) families licensed under a legitimate open-source framework.

Krishna emphasized the significance of true open source for enterprises, stating, "Unlike the Apache license, other open licenses often used are more marketing than actual open source." He stressed that genuine open-source frameworks foster contributions and innovation. “For meaningful contributions, it must be undeniably open source, not merely open source for marketing purposes,” he added.

Advancing Enterprise AI with Watsonx Assistants

While LLMs are vital to generative AI, AI assistants are equally crucial. Companies like Microsoft and Salesforce refer to their assistants as copilot, presenting a more user-friendly approach to adopting generative AI. Rob Thomas, IBM Senior Vice President and Chief Commercial Officer, explained that AI assistants deliver a structured method for enterprises to implement AI solutions.

At Think 2024, IBM is launching three new assistants: the Watsonx code assistant for Java, designed to assist developers in writing Java applications; the Watsonx assistant for Z, aimed at helping organizations manage IBM's mainframe systems; and Watsonx Orchestrate, which empowers businesses to create their own assistants.

Retrieval Augmented Generation (RAG), Vector Databases, and InstructLab

A prevalent deployment pattern for generative AI in enterprises is Retrieval Augmented Generation (RAG). This approach allows AI assistants and chatbots to access real enterprise information not included in their initial training. RAG relies on vector databases or similar support in existing databases. While IBM recognizes the value of vector databases, it does not aim to develop its own, opting instead for strategic integrations.

“Vector databases are trending, with countless options available,” Thomas remarked. “While we’ve integrated with many, we do not feel the need to own that capability.”

Moreover, IBM sees significant potential in InstructLab technology recently unveiled by its Red Hat division, which supports ongoing improvement of models in a streamlined manner.

The Impact of Generative AI on Employment

IBM remains optimistic about the transformative potential of generative AI while acknowledging its implications for society and employment. When questioned about potential job displacement, Krishna provided a clear perspective:

"Productivity gains typically lead to increased business and overall employment. With declining demographic trends in many countries, enhancing capabilities through generative AI is essential for sustaining quality of life and economic growth."