IBM is revisiting artificial intelligence with its watsonx platform, putting it at the heart of its hybrid cloud plans
IBM’s data and artificial intelligence (AI) segment ended the quarter ended September 2023 on a strong note. Its revenue surged 6 per cent year on year to reach $14.8 billion, outpacing the 4.6 per cent growth in total revenue of the company. In the earnings call for the quarter, IBM CEO Arvind Krishna said that clients are increasingly adopting its watsonx AI and data platform along with its hybrid cloud solutions to unlock productivity and operational efficiency.
IBM’s bet on AI is set to grow bigger in the foreseeable future amidst the hoopla surrounding the technology. Its professional services vertical, IBM Consulting, recently expanded its partnership with Amazon Web Services (AWS) wherein the two companies will re-engineer their joint solutions for enterprise clients with generative AI (GenAI), targeting contact centres, cloud value chain and supply chain initially.
At its flagship Think event this year, IBM debuted the watsonx platform, which helps companies leverage the company’s foundation model and other open source versions. This is taking forward the business conversations that it has been having with enterprises on how AI can potentially enhance their efficiency, productivity and growth.
Geeta Gurnani, CTO of IBM Technology for India and South Asia region, states that these conversations usually fall into four buckets. The first is about AI’s evolving capability to interact and converse, since GenAI is more intuitive than its predecessors, offering better first-time right responses.
Next is how it can help to automate IT processes. This was to help professionals break their habit of tasks, especially in robotic process automation, without really looking at its end-to-end process, right the data source, parsing it and looping in the feedback.
The third area is knowledge and insights especially for industries like healthcare, education, or banking that struggle to derive insights from the proverbial mountains of documents.
“We recently concluded a pilot in the insurance sector using Gen AI, since even underwriters found it difficult to go through numerous policy documents manually. Now, they can just extract the insight basis the case and the technology points them in the direction they need to focus,” Gurnani reveals. And the final window for conversation is workflow automation, which is at the heart of digitalisation.
Companies should not consider GenAI as an afterthought but should clearly strategise how to embed the technology in their organisational structure. To do this, Gurnani advises that they start by choosing a platform that isn’t proprietary but is trustworthy in nature and has a targeted approach depending on the digital assets they want to harness.
The reason she advocates this approach is because 11 months into ChatGPT’s launch, most people still mistakenly think that GenAI is a large language model (LLM). AI should go beyond prompt engineering and getting outside responses. It should be about leveraging one’s enterprise data to create value. For example, a financial institution will have security, IT, asset and HR data and these digital assets should be targeted when it comes to business domains and role of AI.
IBM’s ‘Technology Atlas’ notes that six technologies—AI, Data, Automation, Hybrid Cloud, Cybersecurity and Quantum Computing—will continue to play a pivotal role in the entire IT space. Keeping an eye on this trend, it is helping companies marry the watsonx AI and data platform with its hybrid cloud environments. Its cloud-native products are built on Red Hat OpenShift container platform, giving companies the flexibility to decide which model or case can remain on-prem and which ones can move to the public.
Gurnani maintains that this flexibility to deploy wherever one prefers is becoming even more crucial in the Gen AI era, where institutions are not comfortable shifting to the public cloud entirely. The reasons could vary from high investments to lack of control.
“A hybrid cloud strategy is gaining more acceptance amid value-driven conversations,” she notes. “People who are on multi-cloud environments or are either completely on-premise or public cloud are now revisiting their decisions to have a hybrid cloud strategy. Without that, I think there are financial impacts and productivity challenges.”
This is why IBM has its own ethics board that governs the full life cycle of the data it uses to train its own models and checks the different open-source models coming into its platform. This ethics board, which acts like a central body, is responsible for ensuring that anything the company is using in building AI is responsible, trustworthy and ethical.
Ethical use of AI has become a key talking point due to concerns over how data is used and the need for regulations around it. Many companies are reluctant to share the data with various platforms because they feel the LLM models can be used for training.
For instance, while Adobe’s features like Generative Fill in Photoshop and the Firefly text-to-image generator were warmly welcomed by users, a chorus of criticism arose from contributors to Adobe Stock. They argued that the company’s liberal service terms allowed it to utilise its work for training its proprietary AI models without prior knowledge or direct compensation.
The company developed a watermarking tool called content credentials to track when images are edited by AI. Other tech majors like Google and Microsoft are also supporting some form of labelling of AI in their products, including using metadata to distinguish fake from real content.
It all boils down to protecting copyrighted content without stifling innovation and the government has an important role to play while framing a framework. Gurnani recommends, “Precision regulation of AI will take care of transparency and explainability. A one-size-fits-all approach for data won’t work. I am sure the regulators are also looking into it as they are seeing the maturity and adoption of AI, and they should look into the transparency of the systems being built.”
The US and European governments are no longer adopting a soft attitude towards GenAI, an approach they adopted because everybody jumped on the bandwagon without evaluating imminent challenges. Now that they are working out the kinks in the initial adoption challenges, administrative and enterprise stakeholders will build the guardrails to protect sensitive data.
Gurnani reveals that to provide that level of comfort, IBM is launching a few foundational models, like Granite, where it takes full responsibility for any copyright issue. “Innovation does come with some sort of caveats and challenges, but that shouldn’t stop the process,” she points out. “Technology's most crucial role is to improve the time to market for any company and to bridge the time gaps that usually exist, especially for startups who are constantly in a race against time. In this situation, Gen AI can be more generic and artificial intelligence can be a great tool.”
That is where watsonx steps in to help enterprises make the most of the data that resides in their network. Gurnani states that the platform comprises three components. The first element is watsonx.ai, which allows companies to do model management, training and fine-tuning.
The second is watsonx.governance, which helps them govern the process, especially if a model has to adhere to regulatory compliances and takes care of biases, accountability and fairness.
“The third element is watsonx.data, which is suitable for highly transactional reporting. It helps you access data wherever it is residing and allows the enterprise and startup community to query this data in the respective format with a good level of flexibility to train their models,” she explains. “What Netscape did to the internet, GenAI is doing to the whole AI segment.”
In India, this change is most evident in customer care and HR transformation, followed by content generation. The use of GenAI in creating marketing campaigns and designing advertisements has seen an uptick recently.
Automation of IT operations is also seeing an uptick as Gen AI can reduce the number of IT helpdesk tickets. While the technology isn’t new per se, it is now reaching a stage where people are eager to infuse intelligence into the automation process.