Watsonx trustworthy, will help enterprises leverage full potential of AI: VP of IBM India Software Labs

IBM’s recently launched enterprise artificial intelligence (AI) and data platform Watsonx is set to revolutionise the way businesses tap into the vast potential of the technology, said Gaurav Sharma, Vice President of IBM India Software Labs.  

Talking at the AI Innovation Day where IBM showcased Watsonx, Sharma said the platform is based on the three principals and five properties of IBM.  

“The key differentiator with consumer AI is that you can have mistakes in consumer AI, it has to be regulated. Watsonx is trustworthy AI and it is based on the three principles and five properties of IBM. The principles are that we will use AI to augment human intelligence, and not to replace it; the data and insights are of the creator; and there should be transparency and explainability. The five properties are transparency, explainability, robustness, fairness and privacy. Watsonx is based on all these,” Sharma said. 

The platform consists of three components—Watsonx.ai, Watsonx.data and Watsonx.governance. Watsonx.ai is a next-generation enterprise studio providing capabilities for training, validating, tuning, and deploying machine learning and generative AI models powered by foundation models. Watsonx.data enables scalable AI workloads by leveraging the entire data landscape. This purpose-built data store optimises governance, supports AI workloads, and offers easy access, data sharing, and open data formats. Watsonx.governance is a comprehensive toolkit empowering customers to build responsible, transparent, and explainable data and AI workflows. 

Sharma said Watsonx is extensively utilised within IBM and the company has seen substantial productivity enhancements. Watsonx is scheduled to be launched in July on a SaaS model. Initially, it will be available on IBM Cloud and on-premise data centres, with future plans to extend its availability to other cloud platforms.  

The company has been working with partners including SAP Labs, STL Digital and LTI Mindtree to develop its enterprise AI solution for different use cases. Powered by foundation models, Watsonx provides accurate, scalable, and adaptable AI for businesses, Sharma said. 

“A lot of work has gone into enterprise AI. Today, on a foundation model, what we are exposed to is the large language models. IBM has been working very closely on foundation models with various partners including Red Hat on automated code generation; a large pharma company to help scientists predict properties of molecules; and on geospatial tech where we are working with NASA on climate changes and other things. It’s not forward looking, it’s already here. Customers have started using a lot of foundation models, tuning them, doing prompt engineering for their specific use cases,” he said. 

Manish Bhide, IBM Distinguished Engineer and CTO–AI Governance & OpenScale, said IBM Research is building its own foundation models and that models like Watsonx provides a base for supporting multiple uses cases.  

“Foundation models give enterprises a foundation to look at the different use cases they can solve, so that they can hit the ground running as opposed to building something from scratch. With Watsonx, IBM provides enterprises a platform which allows them to use these foundation models, be it open source or IBM’s own foundation models, so that businesses can focus on the five per cent of work that they need to do as opposed to doing everything,” he said.  

Bhide said IBM’s focus and investment on building own foundation models is to ensure that trust is safeguarded. “We are building our own foundation models to cater to the needs of our clients and the reason we are investing on it is because we are very focused on the trust part of it. We are making sure that we are using trusted data while training these foundation models. If you go through our competitors or the open-source models, there is absolutely no guarantee whether the data they use has gone through the right governance (structures),” he added.