Artificial Intelligence

Co-founder and producer at Salvi Media, William Salvi discusses artificial intelligence and big data with Chief Digital Officer of IBM, Nancy Hensley. William and Nancy discuss the daily interaction with artificial intelligence and how we use big data to make better decisions.


About Your Role At IBM

Nancy: My current role is Chief Digital Officer. What that really means is that I focus on making all of our really powerful data and AI solutions much more accessible, much more consumable, easy to use and hopefully we get clients that we've never had before coming to IBM because they are so much more accessible than they were before.

Risk & Reward With Artificial Intelligence

As with any major innovation, there's always going to be the good with the bad, right? If you can think of any huge innovation over time, whether it was being able to travel in space, and then of course the Challenger exploding. There's always risks to anything we do, the challenge is:

Does the risk outweigh the benefits?

And when it comes to artificial intelligence, for sure the risk doesn't outweigh the benefits, because there's so many ways you can benefit from it, in almost any industry you can think of.

But there are risks, right?

And we do have to manage those. And I think we all got a taste of that with what happened with Facebook and with Twitter and the ability for our democracy to be compromised because people are writing some really good algorithms and using chatbots to infiltrate into our social media systems and actually influence people.

So, I think it really comes down to education so people are better at recognizing when there is something risky. It comes down to us as researchers and makers of AI, to help manage bias within the models as well. And to be able to detect when there's something going on with that model that's in production, and that's something we spend a lot of time on today.

Common Uses of AI?

Will: What are some other common, very strong examples of AI in daily life?

Nancy: I mean, your recommendations on Netflix, right?

Your recommendations from Amazon, how you open up your cell phone with your facial recognition, even how you get through airport security, with some of the AI that they have in airports.

They're also using that to scan for people that they might be looking for that shouldn't be in airports so I think it's becoming much more, I don't like the word democratized, but it's starting to become more accessible, and it is becoming more democratized and companies are starting to understand how to leverage it better to not replace people, but to augment and to optimize where they can.

How AI Is Helpful In Ways We Don't Think About?

When you go to a doctor, right? You go to a doctor and they maybe get to read a couple medical journals when in reality there's probably been 200,000 medical journals published that month and they've diagnosed you, and they're coming up with a treatment plan. And they're using all the tools that they have, and the knowledge that they have, and the exposure and the data that they have, but they can only process so much as humans.

With some of the capabilities that we have, we can augment what they know with treatment plans that have worked across the world. And they would have access to that at their fingertips.

So it's not replacing the physician, but it's augmenting their knowledge with a world of information that's available to them, that they couldn't possibly consume.

Bias In AI

That's probably the biggest challenge most clients have is that, they have the ideas, they know what they want to do, they struggle with getting all the data together, they struggle with getting the models into production, and then they struggle with once they're in production, really "How do I make sure that these models are still valid?" and "How do you make sure that I haven't somehow introduced bias with these models?"

Whether it's gender, socio-economic, geographical... Bias exists because the models are created by humans.

Data Effectiveness in Business

Well once you start to use data for insight and you realize how powerful it is. I mean it is like a drug.

Will: Right, yeah. And it's a drug for the consumers too. It's so easy to consume.

Nancy: It's addicting.

Will: I'm going to shift a little bit.

Growth Hackers

Nancy: It's using rapid experimentation across all aspects of the customer journey to find those little tweaks that help you grow.

So, for example, we have on one of our forms to get onto our cloud, sometimes you have to use your credit card and then once you get past a certain capacity, it starts to charge you from a trial usage.

So we were experimenting, if we said "No credit card required," would that increase the amount of trial users? And it did, by 46%.

So that was something in acquisition, right? We created an easy button, so that when clients are in trial they press the upgrade button and it helped increase the downloads,
just taking the friction out of it.

So something at every aspect, whether it's the way they engage, the color of the button, the content, the way they consume the software itself. I mean, there's opportunities at every aspect of the customer journey that we basically have rapid experiments on, all the time going to find those growth spurts.

Will: I just realized I really did make a hard shift from AI to growth hack.

I think in the future as well as like what content, what access do I have to information that anything like a device actually starts to understand you even better.

Next Episode

One North Creative