Why A.I. needs to be open to all

Businesses of all sizes across different sectors are exploring or deploying Artificial Intelligence (A.I.) today more than ever before. This is game-changing for companies, as they look to leverage automation, natural language processing, data-driven decision-making, and a host of other capabilities. They are using A.I. to drive enhanced customer experience, complement their knowledge workers, boost innovation, and become truly agile. In the Covid-19 era, A.I. has had a crucial role to play in improving business resiliency and continuity besides powering personalised customer service.

However, successful A.I. adoption is dependent on a few factors such as moving A.I. from prototype to scale and the availability of skilled talent. To realise A.I. at scale, it needs to be portable across clouds and not restricted to the tech platforms on which the prototype was built. A.I. has been democratised now, which means, it can be made open, available to all who can add, modify, retain, extend the models, and include them in their applications with no code or low code. This is where the ecosystem collaborations come into full force as the partners can extend models and move them to different clouds. They can add bias mitigations, explainability and/or scaling parameters to make the entire A.I. system more robust.

Building the right ecosystem can further A.I. adoption

The App Store is a great example of ecosystem partners coming together to drive success. In today’s world, there is a need to build the foundation platform and make available open Application Programming Interface (API) for partners to add and further grow the base platform. Similarly, open A.I. systems will accelerate the development of applications using available A.I. APIs and the best of technologies. For example, a partner can extend a model for new languages; the other partner could provide the scalable environment for it to run volumes and another partner could develop a scalable chat bot where it could be deployed. Of course, to make it all work the A.I. models should be portable and run on any cloud.

Many times, vendors create intelligent chat bot systems that run on only one cloud. This restricts those clients who have multi-cloud environments and demand flexibility. A.I. services must be available to businesses operating in any cloud environment—public, private or hybrid, as they increasingly create and store data across clouds and data centers. The idea is to enable companies to bring A.I. to their data so they can uncover hitherto undiscovered insights from their clouds and allow them to infuse A.I. into their applications, wherever they are stored.

Another critical aspect in the democratisation of A.I. is fostering open source—the ecosystem must let companies run any open-source technology, framework, or library on their platform. However, resolving issues of vendor lock-in and promoting open source technologies alone may not bolster A.I. adoption as most businesses face the challenge of limited skilled talent. Moreover, within the organisation too, A.I. would be accessible to a select few data scientists and A.I. engineers, leaving out the business domain experts.

The ecosystem partners can bring about democratisation by making A.I. accessible to the broader workforce within the organisation. They should empower domain experts so they can build A.I.-powered applications by leveraging drag-and-drop functionalities in a zero or low code environment. Already, several ecosystem players are implementing initiatives in collaboration with industry bodies and key stakeholders in the private and public sectors to strengthen the ecosystem.

Some initiatives for democratising A.I.

The National Association of Software and Service Companies’ Centre of Excellence on A.I. in association with technology companies provides infrastructure, software, and mentorship to startups so they can build innovative products faster, thereby realizing the vision of Digital India. Further, all the leading Global System Integrators in India have been joining hands with platform companies to develop joint A.I. solutions that are ‘Made in India, for the world’. The Government too is involved in democratization initiatives—for instance, the Central Board of Secondary Education is integrating A.I. curriculum for students of Grades XI and XII across 13 states.

To summarize, collaborations are the key to the democratization of any new technology. Various studies spanning different markets and verticals show how companies using A.I. benefit from higher revenues, improved customer experience, enhanced efficiencies and gain a competitive edge. However, many of them face challenges in deploying A.I.; by collaborating, the ecosystem players can democratize A.I. for all.

Views are personal. Kapoor is IBM Fellow, Director & CTO - AI Apps, IBM India Software Labs and Chahal is Director - IBM Automation & Asia Pacific Cloud Pak Labs, IBM India Software Labs.