Sign up for our newsletter and get the latest big data news and analysis.

Democratizing AI

In this special guest feature, Saranyan Vigraham, Vice President of Engineering at Petuum, believes that as technology creators, we have an inherent bias in solving challenging technical problems. For democratizing AI, it is also important to solve problems around adoption and the last-mile problem of AI delivery. In a manner similar to e-commerce, an entire ecosystem of players will emerge to tackle this challenge down the road. Previously, Saranyan was a VP of engineering at Elementum, where he led their enterprise SaaS platform team that digitized the supply chain for Fortune 500 companies like Starbucks and Johnson & Johnson. Prior to Elementum, he was the head of software engineering at Meta, delivering the industry’s first Augmented Reality platform with natural gesture controls. Saranyan holds a Ph.D in Computer Science with a focus in Machine Learning.

Artificial Intelligence (AI) is a technology accelerator and enabler. When we truly democratize AI, we have the opportunity to cause a massive paradigm shift in how humanity solves problems.

In 2016, TensorFlow showed the power of democratizing AI when a cucumber farmer got excited by the potential of using deep learning for sorting cucumbers. Outside the fact that this put TensorFlow on the map, it showed us a glimpse of what the world would look like when AI is easy to understand, use and access.

There are several examples today of using AI frameworks to solve real world problems. They work better in some areas than others. For instance, image classification and training is a problem that is consistently getting better every day. It is an area of AI that is ripe for democratization.

When we talk about AI democratization, we talk about broad AI, which is a mistake. There are several narrow AI applications where we collectively, as an industry, have made incredible progress. Several of these applications are at a level that is ready for the world to use. Our focus should be to push adoption in these areas while working to solve other areas.

By being aggressive in pushing for adoption, we accomplish two things – we make users familiar with AI applications thereby addressing stigmas surrounding it and we also empower users to perceive AI as tools that augment human decision making capabilities.

The term democratization is very broad by nature. It is also an inclusive term where the target audience tends to be a diverse set. Those of us, who have built products know that it is impossible to design a product for everyone. The best you can do in these scenarios is provide a set of building blocks that are easy to understand and assemble into functional solutions. An AI platform that is capable of supporting a diverse set of users and use-cases is quickly becoming a holy grail in the field. Companies like Google, Microsoft and Petuum are trying to solve this problem to come up with the right set of building blocks that makes adoption of AI easy for some of well understood/mature use cases.

A big challenge today in AI is also the last-mile problem, which essentially focuses on getting technology in the hands of users. Because ultimately, people like farmer Makoto are domain experts, who with the right tools are able to build solutions to improve their lives. Several companies are actively working on these last mile problems by building specialized AI
applications to be accessed from anywhere in the world via cloud or run locally on smart phones. All these efforts to enable efficient narrow AI applications that are accessible at one’s fingertips will go a long way in AI adoption and subsequently democratization.

Platforms like internet and smartphones are critical for AI adoption. The current global internet penetration is 57% in the world (81% in developed countries) and smartphone penetration is 40% (70% in developed countries). It is in our best interest to leverage these platforms to deploy narrow AI solutions, which address focused problems. These channels will propel AI mainstream.

We can look at a case study of Reliance Jio, which disrupted the telecom industry in India in Jio offered free voice, data and messaging applications for the users, quickly garnering adoption and market penetration. It fundamentally reshaped the behavior of users and made data and applications part of their lives. Lowering the barrier to adoption is critical in technology acceleration. The same applies for AI.

As technology creators, we have an inherent bias in solving challenging technical problems. For democratizing AI, it is also important to solve problems around adoption and the last-mile problem of AI delivery. I feel that similar to e-commerce, an entire ecosystem of players will emerge to tackle this challenge down the road.

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: