Microsoft India's Sundar Srinivasan speaks about the "tectonic revolution AI is bringing to human lives" and how his company is helping democratise it.
Artificial Intelligence is one of the biggest things that will impact us humans in the coming years. While there is no doubt that this evolving technology will make life better for humans, there are fears about it triggering job losses as well as apprehensions about a future where AI will be more intelligent than those who created it. Sundar Srinivasan, General Manager, AI Division, Microsoft India spoke to Nandagopal Rajan about the "tectonic revolution AI is bringing to human lives". Edited excerpts:
What is Microsoft's take on AI?
Earlier generations of computing will be defined by humans trying to work with computers, from here on we are entering a new era where the computers are trying to understand human beings better... allowing us to be more human while we leverage the full power of computing. What does this mean? Our ability to understand, communicate and empathise are our key differentiators. For computers it will not be just about understanding sensory inputs, but also what human beings are thinking and their emotional state. Going forward integrating that context and how AI will come into play in terms of assisting humans will be kind of the broad brush holy grail AI can get us.
At Microsoft we fundamentally believe that in this era we need to have computers understand us better, we need to understand emotional intelligence and then democratise and bring these capabilities to everybody and into whatever they are trying to do. We believe we can create the platforms that can empower everyone.
Is AI going to make us dumb as we outsource more skills to the computer?
One way to look at human evolution is to look at how tools made humans more productive. Everyone on these revolutions is about empowering us to shift to a different orbit of productivity. We don't believe humans are being replaced, it is about AI amplifying what humans can do, their ingenuity. It is not either-or, but about working together, kind of doing things way for efficiently and better.
I don't think of it as losing capabilities, it's about simplifying things that take us way too much time so that we are empowered to take on higher order and more ambitious goals. It always is an enabler for us to do more.
What will be the impact on jobs? If people doing mundane jobs lose out what will be the impact on the economy?
AI can help people do tasks more efficiently. It fundamentally does not change the need for a human to do something. For instance, at call centres, AI can eliminate the early triage - like in the case of email now. They are not going to go away, but AI will triage calls and reduce the wait time. That is the model where I think about it as amplifying human beings. AI is never going to be in a situation where it will do things on its own.
How is Microsoft going to integrate AI into what you do?
At a high level our mission is to empower every individual and organisation. While there are lot of examples, broadly we think about this in four pillars: apps, services - like cognitive services on Azure, then Cortana-like agents which bring in ambient and contextual computing on any device, and finally, infrastructure bringing in the power of GPUs running in the cloud to developers who can't afford it.
Is there a culture within Microsoft where everyone has to learn and understand AI since it is something that will be infused into everything?
Every developer will need to understand what AI can do for them. Everyone doesn't have to become an AI or machine learning expert, that is not practical. But everyone does need to be aware how to incorporate AI into whatever they are building. On the other side, we do need a lot more AI developers as a domain, for they need a special set of skills like machine learning, understating big data and working with neural networks etc.
How does India stack up in terms of talent in this field?
We do have academic institutions with people who understand the subject and are producing a lot of the talent. But that talent also needs to be groomed with the right environment. At Microsoft we have that environment and people who can groom this talent.
How are people changing with AI and how are computers learning with AI?
Our handloom weavers are limited by the designs which are created by one master weaver for a few hundred weavers. He takes interesting designs and fits it in a colour scheme and gives it a pixelated form. We thought of bringing in AI to make this simpler by deploying the Ruuh chatbot to create handloom weavable designs. Ruuh engages with a lot of our younger generation folks who are mobile and offers a friend persona, especially when someone is upset and doesn't want to speak to other people. Ruuh is having very interesting conversations. There have been cases where people have chatted over 10 hours with her. The fact that people are able to chat with a bot with that intensity and timelines and scales tells us AI can be a very interesting proposition with the right level of emotional intelligence.
What are the big challenges in getting AI to where is should be?
Now we are very good at narrow AI, for specific and focussed tasks. Those are rule-based AI behaviour. From there it has also moved to perception with sensory things like sight and understand spoken or typed language. We are now going from perception to cognition with context like humans can. That is where we are in the broad sense of AI development. Then AGI or augmented general intelligence is capable of, like with Cortana, to have conversation. It is evolving to doing lots of different things and AGI is about building the intelligence for that at scale and we are at the early stages of that. Then there are possibilities with mixed reality and that is the next decade as we see it.
Do you agree with the analogy that AI is the new electricity?
It is very fundamental. I think AI is a pretty big revolution. How it impacts human race and economy is going to be very fascinating and dramatic. But these are very early stages. It is not as if AI is going to unfold in the next few years, it is going to be a sustained change and is going to take a few decades. We have already gone to through the troughs of disillusionment and is now in the productivity plateau.
Source: Indian Express