5 Key Takeaways from IBM’s Watsonx Day
Everything you need to know about the new reveals and demos from IBM’s watsonx Day
Watsonx Use CasesDr. Kareem Yusuf, senior vice president of product management and growth at IBM, delivered the watsonx Day keynote. Dr. Yusuf spoke about the core use cases of watsonx, especially pertaining to business growth, giving three examples: talent, customer care and application modernization.
He reported that with the watsonx AI platform, businesses could see a 40% improvement in HR productivity, streamlining tasks such as talent acquisition and managing employees and performance. Customer care will also see a boost—for example, using conversational AI to carry out call center tasks and free up employees’ time for high-value tasks. Finally, features like to automated code generation, playbook generation and model tuning could lead to a 30% boost in application modernization and operations productivity.
According to IBM, 80% of business leaders have ethical concerns about AI, such as the explainability of AI-driven decisions or established biases being reinforced by AI. To ease these concerns, Dr. Yusuf said, “To create true business value, generative AI must be tailored to the enterprise.” Watsonx can be customized to do exactly what users need it to do—or not do—in a trusted environment with governance in place.
Watsonx Capabilities DemoIBM also led three demonstrations of watsonx’s capabilities. In the first demo, watsonx was deployed to summarize a 6,000-word conversation transcript into a 150-word paragraph, with a list that included the product, the main topics, the secondary topics and more, resulting in increased productivity.
The second demo showed how watsonx can be used to enhance a virtual chat assistant. In the demo, watsonx used Watson Discovery to pull data and a Q&A API configured to Watson Assistant. It provided answers to users’ questions by accessing existing documentation in a knowledge repository, searching for an answer and outputting it to the user in the chat interface along with relevant documentation.
The third demo explored how watsonx can improve generative searches. Users can type a question into the search bar— for example, asking what support options IBM has for corporate customers. The search will return an answer, a list of relevant IBM webpages, and a sidebar of relevant products and services. Watsonx’s generative AI capabilities expand enterprise search capabilities beyond keyword-based document retrieval, using in-depth data sources to get users to the information they need faster.
The Paradigm Shift of Generative AIIn another session, Sriram Raghavan, vice president of IBM Research AI, spoke about how the generative AI used by watsonx can help businesses increase scalability and build trust. “We are really in the fourth generation of AI,” Raghavan said. This era of AI is centered on the ability of foundation models to learn from large quantities of data, adapt quickly to many tasks, accelerate enterprise adoption and do self-supervision at scale.
IBM is building watsonx around “the ability to accelerate enterprise adoption way beyond what was possible with the previous era of deep learning,” according to Raghavan. In the current era, foundation models can self-supervise at scale. This means that AI has the ability to intake raw, unlabeled data and then generate labels for the data itself. This streamlined process will make handling massive amounts of data far more efficient, encouraging large-scale enterprise adoption.
Watsonx.data Deep DiveIn his session, Joshua Kim, product manager at IBM, spoke about the power of watsonx.data and how it can allow users to scale AI workloads. He revealed that stored data is set to grow more than 250% over the next five years. However, 82% of enterprises still use data silos, which are ineffective when data needs to be accessed across an entire organization.
A data lakehouse is the preferred solution for many modern organizations, taking the low-performance, flexible storage options of data lakes and the high-performance, structured storage options of data warehouses and combining them to support structured or unstructured data at a performance level fit to purpose.
Watsonx.data builds upon the data lakehouse architecture in many ways, such as the ease of data consumption. Users can access their organization’s data and the lakehouse uses a virtual assistant to take in questions, scan its store of data and provide an answer.
Open-Source Foundation of WatsonxOpen-source technologies are the foundation of watsonx. For example, watsonx.data utilizes the Presto open-source SQL query engine instead of building, deploying and maintaining a proprietary IBM engine. IBM also uses CodeFlare as an open-source method of developing models in an automated, simplified way; additional open-source technologies like Ray and PyTorch are deployed with CodeFlare, showcasing how multiple open-source technologies work in unison to bolster watsonx’s capabilities.
Watsonx uses several open-source technologies to tune and serve foundation models. Hugging Face is one example, which contains hundreds of thousands of open-source models and tens of thousands of open-source data sets that allow for unrestricted access by watsonx customers. According to Brad Topol, IBM distinguished engineer, director of open technologies and chief developer advocate, IBM has a “commitment to delivering clients an open ecosystem approach and allowing them to select and use the best models for their business.”
These are just highlights from IBM’s watsonx Day. Learn more about the watsonx suite here and watch all the on-demand sessions here.
About the author
Christopher Garritty is a content marketing intern at MSPC.
See more by Christopher Garritty