AI and analytics Gearing for computing demands of next-gen AI apps

There was a session on AI and analytics -- gearing for computing demands of next-gen AI apps, at the India Mobile Congress 2021. The participants were: Xavier Kuriyan, Director Solutions and Alliances, Global Compute and Network, India, Dell Technologies, Bill Wright, Head of AI/ML and Intelligent Edge, Global Industries and Accounts, Red Hat, and Dr. Shailesh Kumar, Chief Data Scientist & Head AI COE, Reliance Jio. Sachin Arora, Partner, KPMG, was the moderator.

Dr. Shailesh Kumar, Reliance Jio, said we have had algorithmic growth of AI, etc. We are now taking stock and move forward. We need to revisit fundamentals. Deep learning is everywhere. In the next 20 years, we are going to re-invent AI. They will add social use cases, such as education, agriculture, etc. Till it reaches the masses, AI will not be counted as a success. Lot of work is happening in finding the right use cases. In healthcare, we are developing heath report cards. Accurate and timely diagnosis is important, and how can AI help. Genomics is another area. For farmers, we need answers regarding crops, weather, etc., using AI. In education, we need to personalize education for every child using AI.

Today, we have mastered the art of deploying large models. We now need to have individual APIs. We need to re-imagine the OS and AI. We also need country-specific AI. Lot of India-specific AI will develop. Localization and customization will be the next big thing. There will also be democratization. We need a platform to contribute AI APIs. We are very excited to Jio.

Sachin Arora asked: do we have the talent skills among us to do all this? Dr. Sailesh Kumar said the skillsets for data scientists will transform, and they will be AI architects for our ecosystem. He/she needs to put all building blocks together. We will now need solution thinkers. We need to give training to the next generation to be solution builders. We need to combine data science and domain. We need to develop new skills. Data scientists will be transformed into system thinkers. AI architect is the new role, replacing data scientists.

New scheme of things
Next, Arora inquired about the role Dell is playing in the new scheme of things at AI. How is it impacting society? Xavier Kuriyan, Dell Technologies, said AI is already on the path. It is being taken in smaller strides. Eg., can we take mistakes of radiologist out of the system? Data required for doing these are vast. In a factory, it is important that people walk in with a mindset of working. Lot of data has to be stored, and lot of referencing has to happen.

Within India, lot of data has to be collated. Dell is putting such systems together. We have systems to collect, store, protect, and process data. We have small ISVs based out of residential nodes from India. On app and algorithm side, lot of work is going on. The world is moving to faster data storage. Today, there are multiple options.

Arora asked how is the chip shortage impacting AI? Kuriyan said there is a global shortage of chips. We have managed to circumvent most of the problems. Today, it is important to say: what do I need to get this function to work? It may use run-of-the-mill CPUs. We have not seen shortage of general-purpose CPUs.

Next, what role can Dell play with CSPs to democratize AI? He said, cloud was the first step for democratizing IT. Systems to link cloud environments has been established. The next step is to create data at places where you can't think about right now. The amount of data generated by a car is more than a phone. Self-driving cars generate terabytes of data. The next step would be to have small, distributed edge. Imagine, you can run AI at a store for a short period of time. Cloud and AI need to work together, and create the architecture for the future. Telecom service providers (TSPs) have a big role in creating the architecture, so that compute at the edge exists.

New models
Bill Wright, Red Hat, added that pre-ponderance of AI models has created new models. We are deep in the guts of wiring data centers and networks. We have seen distribution of elements of AI across enterprises. This is creating new network opportunities. You can create single-train image of AI. Red Hat is working on this right now. We can have an enterprise neural system. Ravi Sinha at Jio is part of the project.

The advancement in chips has been massive. It has also opened the door to more creativity. CSPs are looking at the opportunity with hungry eyes. Every 20 years, you get lot of innovation, and this is one of the times.

India has made great leaps in advancement. It is at the forefront of innovation. Subnetworks across countries are really the foundation for a global neural network. India has vast networks, and so have others. We have to tie those together, look for frameworks that can protect the data, and make them work for global good.

Arora next asked the question as to how is quantum computing going to impact India? Xavier Kuriyan, Dell Technologies, said work is going on in this area. Some architectures are now disaggregating. There is enough for certain problems. Quantum computing may not pop up all around soon, but we need to start preparing for such eventualities. Time to implementation has massively shrunk. Earlier, typical deployment times were longer. Today, it is: build an app and role it out. Ideas are now being tried.

Finally, how can the economy react to AI with creators and builders coming together? Dr. Shailesh Kumar, Reliance Jio, said every time something new is invented, new things will happen. With AI, we may see things being developed in data-driven manner. Eg., supply chains can get better. Precision and pro-activity will also happen. AI can also predict something can happen with crops. Optimization will happen at the societal level. AI will lead to a labeling economy, democratize modelling, etc. APIs may even get used by several consumers. It will bring in efficiency and reduce costs. As humans, we need to reskill at the same pace. We need the right skillsets to build an AI-first economy. AI will be an engine of economic growth in many ways.