Published on October 18, 2023, 11:16 am
With the rise of generative artificial intelligence (AI), it’s important not to overlook the pioneering technology that paved the way for human-machine interaction: IBM’s Watson. Originally introduced in 2011 with the goal of winning the game show “Jeopardy,” Watson was an impressive technology capable of answering questions posed in natural language.
Fast forward to today, and IBM has once again made waves in the AI market with its latest offering: watsonx. This new platform is IBM’s response to the growing demand for generative AI and consists of three core modules designed to accelerate AI and data processes for maximum efficiency and innovation.
One of the key components of watsonx is watsonx.ai, which focuses on training, fine-tuning, and deploying generative AI models. It also includes foundation models and machine learning functions. Another module, watsonx.data, is a data store specifically designed to handle scalable AI workloads. Finally, there’s watsonx.governance, an AI governance tool set to be released later this year.
IBM has recently announced several enhancements to the watsonx suite. The first notable update is the introduction of new large language models (LLMs) from its Granite model series into watsonx.ai. These models support a range of natural language processing tasks such as summarization and content generation. They are highly efficient with 13 billion parameters and have been tailored to meet the specific needs of enterprises.
In addition to LLMs, IBM has also added Meta’s Llama 2-chat 70 billion parameter model and StarCoder LLM for code generation. Moreover, IBM will soon release its Tuning Studio feature that allows users to fine-tune foundational models using proprietary data. There will also be a synthetic data generator for low-risk AI model training.
Watsonx.data is getting an upgrade as well. Planned for release later this year, it will integrate generative AI capabilities from watsonx.ai. This integration will allow users to access and refine data for AI use cases through a self-service interface using natural language. Moreover, Watsonx.data will feature a vector database to support retrieval augmented generation (RAG) use cases, enabling LLMs to retrieve facts from external sources in real-time.
As for watsonx.governance, IBM will be launching a tech preview of the platform soon. This governance tool aims to provide approval processes for AI workflows, ensuring human oversight, and automatically documenting foundation model details, metrics, and risk governance capabilities through accessible dashboards.
IBM’s focus on updating and enhancing each element of the watsonx platform demonstrates its commitment to delivering a comprehensive suite of tools that support AI-driven operations in enterprises. By addressing the need for an AI-friendly data store, scalable infrastructure, and robust governance tools tailored to LLM monitoring, IBM is showcasing the maturity of its technology.
With these enhancements, IBM’s watsonx is positioned to compete with other generative AI offerings in the market. However, what sets it apart is its holistic approach that caters to the unique needs of individual enterprises. By providing AI development tools supported by multiple LLMs and an integrated ecosystem of data management and governance solutions, IBM empowers organizations with end-to-end generative AI development capabilities.
In conclusion, IBM’s watsonx enhancements mark another significant step towards advancing generative AI technology. With its history of innovation through Watson and its comprehensive suite of tools focused on AI, data management, and governance, IBM is well-equipped to meet the evolving needs of enterprises as they embrace generative AI in their operations.