Thought Leaders in Artificial Intelligence: Josh Sutton, Data & Artificial Intelligence Global Head at Publicis.Sapient (Part 6)

Sramana Mitra: Very good. Very helpful interview. I’m finding it helpful to calibrate. I know all of these people but things are evolving. I know Michael who’s the CEO of Cycorp.

Josh Sutton: Michael is the CEO of Lucid. Doug and Steven are the CEOs now. There are two other use case that are probably worth mentioning quickly. One is around the entire virtual assistant space. I’m explicitly saying virtual assistant chat bots. I think chat boxes are a baby step forward. A lot of people are making mistakes now in trying to invest in things that have limited upside. Chat bots are simple.

To me, the step forward to virtual assistants is much more around the ability to ask clarifying questions, understand the ancillary drivers that a person might have and provide recommendations in the same way that a good sales person wouldn’t just ask you what you want to buy. Then moving forward into the expert assistant. We spend a lot of time building platforms for companies. This ranges from very large purchases like automobiles down to cosmetics. Moving from that chat bot model to a virtual assistant model that actually tries to engage the customer rather than just respond to direct requests is an area where I see a lot of things going. I think chat bots are going to hit a wall that they aren’t quickly able to pass because they aren’t architecting for the future.

We’re doing that with some major theme parks and some major auto industries. The last use case is within the finance space. We’ve been working with a few different organizations on what it means to become a cognitive organization and be able to systemically mine their data that they have in a way that an analyst would. This is moving into that expert assistant category.

If you think of a large organization, they have hundreds of thousands of proprietary documents that they try to pull information out of to provide recommendations to their clients. We’ve been working with them to build some platforms that enable them to do that systemically. If you think about what Google does for search and Internet, we’ve been using a number of AI tools and a combination of machine learning and natural language tools to enable them to conceptually query all of their internal documents in a way that reduces the time-to-respond for information from days down to minutes by putting the full power in the collective intelligence for the organization.

Sramana Mitra: When it comes to expert systems, the use case is slightly different. What platforms are the best for creating expert systems?

Josh Sutton: I don’t think there is a single platform that is the best. What we’ve typically done in those scenarios is combine a few different capabilities. What I think needs to go into those is a good natural language interaction platform with very robust frameworks around your natural language interaction platforms.

Sramana Mitra: Which ones do you use for that purpose?

Josh Sutton: We use both and we do some stuff with Microsoft as well. Given the breadth of customers that we work with, certain people have biases that we’ve chosen to make sure we have capabilities across all of them. You also need machine learning engine to fine tune your recommendations. You need some type of ontology expansion capability. This falls into your natural language ontological expansion tools. There’s a few out there like Watson and Luminoso that can do that. There are some open-source stuff from Stanford that’s not bad.

It becomes very critical as you look at ingesting all the content and being able to systemically match concepts and query. Then you have your deductive and and reasoning engines. Depending on the use case, there’re a number of different pieces that you can leverage from some of your major players as well as some of your boutique players.

This segment is part 6 in the series : Thought Leaders in Artificial Intelligence: Josh Sutton, Data & Artificial Intelligence Global Head at Publicis.Sapient
1 2 3 4 5 6 7