Seth Dobrin PhD, IBM’s Global Chief AI Officer, shares how he’s working to promote responsible AI use at IBM, his vision for AI at IBM and why 2022 is the year AI ‘gets real’
What would you say are your top professional achievements of the past 12-24 months?
If you look at what I’ve been focused on for the last 12-24 months, it’s been around promoting this concept of human-centered AI, and there are a few parts to human-centered AI that are all intertwined.
The first is, when we build AI, we need to think about who’s going to be impacted by it and who’s going to be using it. The second is, why is AI being built in the first place? So, what problem is it trying to solve? How is it attempting to make someone’s life better, cheaper or faster?
When you have those two things in mind, it’s easier to put trust and responsibility front and center. Then, it’s much easier to understand things like, what type of biases should I be worried about? What type of explanations are required for responsible use? And how much do I need to worry about preserving the privacy of the data, the individuals and the model endpoints? So, it gives you a sense of how robust it needs to be and how privacy-preserving it needs to be.
I’ve been focused on that for the vast majority of my career. Over the course of the last 12 months, we’ve taken a framework that I’ve been working on for the last decade or so and we’ve systematized it. It’s part of an Enterprise Design Thinking for Data and AI framework that’s tied to an overarching AI strategy methodology we’ve developed. That connects the human to the business strategy, to the AI strategy, to the work that the data science and AI teams and data teams are ultimately doing.
What’s your vision for AI at IBM? And how have you been working towards that vision since you became the company’s Global Chief AI Officer in February 2021?
AI is fundamental to IBM’s overarching strategy. IBM’s strategy is now singularly focused on hybrid cloud and AI. Our hybrid cloud strategy is pretty much set, focused around Red Hat and OpenShift. But our AI strategy, up until about 12 months ago, was pretty diffuse.
Over the last 12 months, we’ve really focused IBM as a company on three things that revolve around making Watson the AI for business. The first thing that’s required to do that is fully understanding the language of business. The second is the ability to automate processes and workflows using AI. And the third is trust. So, building trustworthy AI in and of itself, as well as infusing trust into language and automation.
From there, it’s a matter of delivering these capabilities anywhere. That firmly ties it back to our hybrid cloud strategy, meaning that we want to enable our customers, the users of our AI, to build and deploy their models anywhere with trust.
How do you expect the AI space to evolve in 2022?
I think 2022 is the year that AI starts to get real. Like, really real.
First, this whole concept of human centered AI is going get real.
Second, integrating trust into AI is going to get real.
Third, what we call large parameter language models are going start getting real and finding real production uses.
Fourth. we’re finally getting to the point where transformer models or transfer learning are becoming real, meaning that, similar to language models, you train base models on large corpuses of data, and then you need smaller data as a company to implement those.
Fifth, we’re starting to see AI reasoning becoming real. So, things like neuro-symbolic learning and logical neural nets.
And finally, we’re going to start seeing edge AI become a reality.
What’s next for IBM’s AI strategy? And what are the key things you want to achieve in the coming year?
We’re multiple different businesses. We’re a research division. We’re a consulting business. We’re a software business, and we’re an infrastructure business. So, one of my roles has been to pull IBM together around some specific areas, where I feel that IBM is uniquely qualified or positioned to help our customers.
The first place we brought the company together was around trustworthy AI. That’s where we focused in 2021. That will continue. And the flywheel effect is taking hold and that’s going on its own.
The next space we’re looking at is around edge AI. So, how do we bring to bear the capabilities of IBM, meaning we’re uniquely positioned to help our customers, around operating AI at the edge?
From the perspective of our AI consultancy business, we are reimagining how we interact with consultancy customers fundamentally differently, to help them drive outcomes with unique value propositions.