Podcast transcript

Make an Impact with AI

Interviewer: Today we’re talking with Mike Hollinger, Distinguished Engineer at IBM’s Sustainability
Software Group. Mike tell me, how did you get into the field of AI what interested you early on?

Mike: About five or six years ago, we built a great software stack, a great hardware stack in IBM for AI
and for scientific research we actually built the number one and number two supercomputers in the
world. Some in Sierra. Along the way, we needed to build a software system to go with that and I happen
to be a little bit in the right place but also having worked with ISV's and software vendors before, a little
bit the right time to be able to step up and offer to help build out that ecosystem. That meant that my
team was tasked with building showpieces for our API stack. We started with computer vision and we
intended to go build a bunch of other things. And it turned out that the vision application had a full
business by itself, which is what brought me into the Maximo business and the sustainability area.

Interviewer: So Mike, when you begin thinking about working in the field of AI, were there any people in
particular who helped you make connections and choices that helped you get started?

Mike: There were, I have what I consider my board, the people that I can turn to just ask questions
whether you want to call them mentors or coaches or any of that, but I actually asked a couple of my my
board members in quotes what do you think about this? There's all this stuff going on and it's always
interesting, when I ask the same question to a bunch of different people, when they all say the same
thing that's, you know, slam dunk easy. When they all say a different answer, that's the interesting thing.
So when we were getting into this area, when there was a bunch of innovation on those starting to
happen around 2015, 2016, 2017, the future for exactly what was hype and what was real was not clear.
And my board disagreed about is how important stuff was. I chose to get into it and I haven't looked
back since then. It's been a lot of fun.

Interviewer: What excites you most about your work in AI and sustainability?

Mike: With our AI platforms, it's really easy to get hung up on the jargon and the lingo and the really
neat, weird problems you can solve or the things that you could maybe solve but maybe it's only going to
work some of the time. What I love about this area is that there are constant advances. There are things
that from month to month, from week to week, we discover as an industry something new, some new
way to apply what was old now make it new again and to keep on advancing in computing. There's this
idea of Moore’s law that dates back to the original transistors back in the 60s, but the premise is that
every 18 months the cost of computing drops in half or given a piece of workload that's what's driven
innovations, driven the world economy for the last half century or more. And we're seeing that that
hardware advance is not delivering quite as much of that. And now it's coming from software and
applications like artificial intelligence for detecting anomalies and sensor data or for detecting misfields
or something on a manufacturing line. Those are all powered by really clever software stack.

Interviewer: In what ways do you think your work in AI and sustainability might help change the world?

Mike: The reason why I do what I do, the reason why I choose to work at IBM and work in the space is
that I know it's changing the world. I know that there are vehicles out there today that are some of the
top selling vehicles that are electric vehicles that use our technology in their manufacturing process to
help maintain quality, to help eliminate waste, to help keep costs for our customers down. I know that
there are examples where we are using our software to drive sustainable water in certain cities in the
world. I know there are examples where we are driving really neat innovations or just spaces that people
are going out into the world enjoying, and I love that. Given the growth of what I call applied AI which is
getting something off of a notebook or off a whiteboard or out of the lab and putting it into someone's
hands, we're now able to take these really clever ideas and put them into production in a way that serves
not just the data scientist or not just the end user but serves that entire organization. And that's
transformative to me because that's helping our clients really drive innovation for their clients or their
customers.

Interviewer: Mike, what advice would you give somebody wanting to pursue a career in AI and
sustainability?

Mike: The space is one that's really advancing and transforming, but at the same time is a space that's
been around for quite a long time. So, there's a really interesting opportunity for individuals that
understand multiple disciplines to be able to drive innovation. So, if you happen to have hands on
experience in a given industry, whether that's utilities, or renewables, or water, or manufacturing, or
anything right. If you have that experience and you know the types of problems or the types of issues
that come up and you are able to understand the new tools that are now in your toolbox, things like
transformer models that power something like GPT or IBM's transformer systems or foundation models,
which are now becoming a huge driver in the industry. If you know what these things are capable of and
you know the little tricks that come out like hey, if I do this then I can get a better result or if I change the
prompt, which is even an idea, if you know the prompt even exists right you can then start to map that to
the experience that you have, the types of problems that you that you've seen, and you can start
applying the right kind of tool to the right kind of issue. I sometimes see as a kind of a counterexample,
people that pick up a new trick, pick up a new skill, and they immediately turn around and try to apply
that to everything under the sun. That sometimes leads to success, but sometimes leads to these really
weird outcomes that are kind of hard to quantify. It's kind of the example of you know someone holding
a hammer everything looks like a nail right. If you can map the capability and really understand the AI
platform to a real problem and really understand the underlying problem, you can choose the right
techniques, choose the right systems, and understand what kind of data you might need to have and
then from there choose how you want to take action. The space of AI in quotes this is a huge, huge, huge
space that even talking through it with people they'll say well let's just use AI and from 10 different
people will mean 10 different things. And being able to come to that specific agreement of exactly what
we're choosing to do whether it's applying some anomaly model for time series data that might power
our predictive maintenance software or a convolutional neural network or anomaly detection, which
drives our manufacturing qualities thing for visual inspection or anything else right. Those capabilities
are something that you need to have a good lay of the land for and if you've got that combined with the
understanding of the problem space you'll do really well.

Interviewer: Thanks, Mike, for sharing your ideas with us. We've been talking today with Mike Hollinger,
distinguished engineer in the IBM sustainability software group.