Artificial Intelligence

These are the practical uses for artificial intelligence in business

In order to plug AI into your existing workflow you must first understand and organize master data sets, says Schneider Electric Chief Digital Officer Herve Coureil.

Schneider Electric Chief Digital Officer Herve Coureil sat down with TechRepulic's Dan Patterson and talked about practical uses for AI in business. The following is an edited transcript of the interview.

Dan Patterson: This may sound like an elementary question. Practical uses. Right? How are we seeing not just business use AI now? I think we can all kind of point to some examples, but give me the next 18 to 36 months and help us understand, should companies, should enterprise companies, build, buy, or innovate?

Herve Coureil: That's a good one. My own take on that is that, I mean, make or buy are the usual two polar opposite, but actually, there are many shades of partnership in the middle that actually are really interesting. We are spending quite a bit of time on all those shades in the middle, so it could be crowdsourcing, it could be working with data science platforms or startups, et cetera. But then really co-developing, co-innovating, et cetera. So we try to move a bit away from this make versus buy and try to be a little bit more creative in the middle.

There's tons of exciting things we are working on, actually, from an AI standpoint. We have a team that's supercharged. We are actually developing more and more algorithms. We're also starting to think also about the governance of the algorithm, how you expose them for API, how you make sure that if somebody already exists you reuse it, you don't reinvent something else, et cetera. So there's a notion of governance that is also kicking in. And then machine learning, of course, is a big deal, but we're also looking at other things like applying, for instance, via Asian network to understand a network of data in the building and being able, for instance, to help a human being tagging master data, right. Making the construction of your data model easier. So that would be some of the other things we are working on. That might be slightly more abstract, but that we hope to bring closer to the market and really having solving practical issues. Because if you have a large building, or the identification of your master that have all your sensors, that takes a lot of time. So if you can accelerate that...

SEE: Cybersecurity strategy research: Common tactics, issues with implementation, and effectiveness (Tech Pro Research)

Patterson: Master data management kind of goes hand in hand with AI.

Coureil: Totally. I knew somebody who told me always, and I really like it, there is not AI as in artificial intelligence without IA as information architecture. And that's so right.

Patterson: Okay, last question here. Let's talk about the Cloud, and not the Cloud in the speculative future, let's talk about how the Cloud and how AI helps companies be more efficient. Let's talk specifically. We have some Clouds, Azure, we have the Amazon Cloud, we have the Google Cloud. So, how is AI benefiting business now using those public Clouds? And where are we going in the short term?

Coureil: So that's actually I see, if you will, three levels very, very quickly. The first one would be you buying a software as a service that has some net AV/AI capabilities. Case in point, we're a big user of the Salesforce. Salesforce comes with Einstein Analytics so you may have as part of softwares and service package some AI capabilities that you're just going to use, I would say almost out of the box, right.

There's a second category which is edible US, Azure, which basically you have an application and you want to modernize the interface to that application. You want to make it voice enabled. You want to bring vision. You want to bring a conversational interface. So then, of course, what those Cloud providers are offering is amazing because with API et cetera you can suddenly add a layer of intelligent interface to an existing application, right. That would be the second use case.

SEE: Research: Defenses, response plans, and greatest concerns about cybersecurity in an IoT and mobile world (Tech Pro Research)

The third use case is back to the earlier point, you're looking at a particular problem it can be... So, you want to really develop a specific algorithm. So we, of course, working with the Cloud, then those as well, but not as just consuming an API more as co-developing with them or with partners using some machine learning models most of the time that are available or libraries. A framework trying to solve some of those particular problems. But, some of those particular problems it's actually... You pointed it out, the first thing is do you have access to Vidalia? Do you understand Vidalia? So it's really understanding the architecture, being able to tag the master data et cetera. And then are you able, on the other side of the algorithm, to plug this AI into an existing workflow. Because you can have the best AI and the best analytics in the world but if you can't plug it in an actual workflow so that something happens on the other hand, well it remains a nice idea. But you're lacking practical operability.

Dan Patterson: Brilliant stuff. I lied. I have one more question for you.

You're a thought leader in you industry, you're making decisions. So you must be consuming content as well.

Herve Coureil: Oh yes.

Dan Patterson: Who are you reading, and who are the thought leaders in the industry? And who can, if I'm watching this who can I follow to learn from and to really help add nuance to the AI conversation?

Herve Coureil: Well, I spent quite some time to curate a good Twitter feed.

Dan Patterson: I know you do.

Herve Coureil: So, of course you would have all the traditional suspect. But I do like some, I think, of a big VC firms. Speak of almost a Sicoya or speak of an Orabets, they have awesome content actually on AI. So I really am following avidly some of the main partners here, and some of their content. Because I think they really develop interesting view points of the market. Then, of course, I'm a big reader. There's a couple of books that I read recently on developing that I think are probably interesting.

There's one from Pedro Domingo, 'The Master Algorithm' that's pretty amazing. And another one called 'The Book of Why' about causation from Judea Pearl. So that's two super book that are kind of a good way to get started about thinking on AI.

Also see

20180713herve4dan.jpg

About Dan Patterson

Dan is a Senior Writer for TechRepublic. He covers cybersecurity and the intersection of technology, politics and government.

Editor's Picks

Free Newsletters, In your Inbox