Once the shelter-in-place orders lift, how will things be different? Can the use of artificial intelligence help us re-start and re-build in quicker and more efficient ways? In this interview, we spend some time with Deborah Leff, Industry CTO for Data Science and AI at IBM, about these and other issues related to COVID 19 and our response to it.
Amanda Buckholt (00:15):
Coming live from Dallas, Texas is Amanda from Cresco International. You know, they say everything’s bigger in Texas and that includes the stars we bring on our show. Here today we have with us Deborah Leff, IBM’s Global Leader and Industry CTO for Data Science and AI. On the Cresco side we have our VP of Services Kenny Mobley joining us to discuss AI in a post pandemic world and shed some light on how it can be utilized for a speedier recovery. So Deborah, that’s a pretty interesting title that you have. Can you expand a little bit more on what you do at IBM?
Deborah Leff (00:50):
Certainly. So within IBM ‘s data and AI business, we have added a group of industry CTOs and our role is to work with specific segments of the market all around the globe and to help them as they establish their strategic roadmap of AI and machine learning initiatives that they would like to attack to optimize their business.
Kenny Mobley (01:14):
So Deborah, what are the type things that they were looking to use AI for prior to the pandemic? What type of things were they concentrating on the most?
Deborah Leff (01:25):
So I think we’ve been in this march towards more optimized operations and enhanced customer experience. One of the greatest things that machine learning models allow us to do is to really learn from all of the data and all of the interactions that we can collect now and have access to now that allow all types of retail, consumer product , travel companies and across all industries to drive a very personalized experience. So within the industries that I specifically focused on, which are all consumer related and travel related, there was an absolute drive towards investing in anything that impacted the customer experience to make that more personalized and frictionless as well as drive operational efficiency, efficiency within the organization.
Kenny Mobley (02:21):
So I would guess with all the things that have been happening now, there’s been major disruptions in both the consumer purchase side and in travel specifically, I’m guessing they’re going to have to look at it a lot differently post pandemic.
Deborah Leff (02:35):
Well, I mean, I can tell you as someone who has serviced markets for a long time and I have relationships with some of the leading, you know, companies around the globe, it is heartbreaking to see how they are being impacted right now. And it’s almost unconscionable to even imagine a scenario where retailers are shuttered and have to figure out how to continue to service customers. What’s happening with grocery is completely different in that there’s the stress on their supply chain. I mean, I think we’re all feeling that pinch trying to scramble to get cleaning products and paper towels and toilet paper and seeing all of the break points in the supply chain being just exposed and laid out there. For the travel industry, it’s just heartbreaking to see hotels shuttered and airline routes, you know, roll back as it’s just not safe for people to travel and to move around the country, let alone the world.
Deborah Leff (03:33):
So, you know, we’re living at such an unprecedented time and it’s unbelievable to see the impact. So many things that people were working on have been absolutely halted and there’s no doubt there’ll be a new prioritization of, of efforts. Right now we’re in the throes of the crisis. I imagine there’ll be some leveling off of as we see the daily weight of new confirmed cases start to level and then we’ve got, god-willing, taper and then there’ll be a point in time that we can start thinking about recovery after the fact. But I think right now as the number of cases in the United States are growing exponentially, I think right now we’re still in the early days of this crisis and it’s all hands on deck to support employees and customers and do everything we can to minimize the disruption. But doesn’t that feel like a horrible understatement, as our world has been upended so we’re as disrupted as we could be.
Kenny Mobley (04:43):
It’s great to see all these commercial enterprises coming together to help because it is a humanitarian crisis at its core. But I can see how prior to all of this happening, you’re talking about personalizing to a customer and it being very much about optimizing, trying to get the last bit out of something. Whereas post pandemic, they may be going back to a lot of fundamentals, like you say, supply chain, human resources, things that they haven’t worried about at that scale for a long time. Do you still see a place for technologies like AI and helping on those fundamentals or is AI really only good when it comes to things like optimization?
Deborah Leff (05:26):
Oh, I think we’re going to look back at this time period and all of our lives and I feel like there’s going to be a lot of pivots that we can look backwards and say, you know, this was because of the COVID-19 crisis. I do think that, you know, obviously the things that people want to do and that support growth will be important. But I think there were a lot of competing priorities and I’ve seen the way companies have approached AI. So if you just look at the way AI has evolved over time, it really wasn’t until the last handful of years that this even became a real possibility for most companies. I don’t think we started talking about AI in a commercial sense before
Deborah Leff (06:16):
Watson won a game of jeopardy. That was in 2011 and it wasn’t until we started getting comfortable with this influx of big data, which is so important to training models that we started to experience the maturity of cloud computing. So we have this unlimited compute power to actually process all of this data. And then that led to all of these advances in machine learning and artificial intelligence. And before, you know, early-mid 2011, 2015, most of most of AI and ML was out of reach for most companies. It was either an academic pursuit or perhaps there were some very specialized resources and operations research and not every company had invested in operation research teams. So certain certain industries have been more advanced I would say. We see more advancements in oil and gas and in some engineering we see lots of adoption of IOT devices and what all of that data can do.
Deborah Leff (07:24):
But there’s lots of industries where it was something they wanted to do. As all the press and the analysts started really talking about what could be done, that really raised the visibility into adopting ML. But a lot of companies were really just starting out. And the successes that we’ve been seeing are relatively recent and even though there’s been a lot of focus and investment, there’s also been a lot of really not understanding how to get to AI at scale. Harvard Business Review published an article in August that talks about the fact that very few companies are really structured with what they need even achieve AI at scale. The number was like 8%. It’s really, really low. There’s lots of articles written about it, how companies are struggling, that they get stuck in experimentation – things that work in a lab don’t necessarily work in
Deborah Leff (08:21):
production. And that’s because we never did that before. This is the way that we’re creating models and infusing them into applications and mobile apps and websites that’s all very, very new technology. And I think we’re going to look back at this point in time and we’re going to say, but this pandemic was such an incredible catalyst for companies reevaluating how they need to attack these projects and making sure they get all the way through into successful. And whereas it wasn’t that the sense of urgency, like these are cool things we could do. These were interesting things, but they weren’t like it. We wanted to remain competitive. We were concerned about disruption, but they weren’t urgent. And I think we’re going to look back at this time period and say that, wow, we really accelerated getting to AI at scale because it became urgent and necessary.
Deborah Leff (09:23):
And if you look at what’s been happening with, you know, just remote work and distance learning there, there are so many companies that for whatever reason just haven’t been embracing those things until now. We have lots of office workers that, you know, work for CPG companies or in banking or in other industries that for sure that just never worked remote and now that’s all they have. And that creates, you know, mud because necessity definitely does become the mother of invention. I remember last year once my daughter had the flu and as she was recovering, she was really well enough to go to a class online. I wouldn’t have sent her to school because she could have had a relapse or infected others. But it was really a shame that she was home watching Netflix all day, which really could have been a distance learning. And there was no embracing of that. And now what’s happening around the globe, how overnight almost now all of these institutions that have embraced distance learning.
Kenny Mobley (10:34):
Yeah. Cause you have to think that’s going to be an overall just culturally positive because before there was this idea that you perhaps weren’t serious about what you’re doing if you didn’t push through maybe feeling a little sick or go to the office anyway. Part of it was because there weren’t very good structures set up for you to stay home and part of it was cultural. But perhaps having those structures set up will reduce some of the cultural effects of it and say “Hey, if you’re feeling ill today, don’t go into the office and take the chance of getting other people sick” and this will make people be a little more mindful of the effects of contagion and other things. Having those structures set up will turn out to be helpful. Do you think that companies that are trying to kind of get back on their feet will look to things like AI as an accelerator to what they were doing, whereas before they were trying to optimize, but now they’re thinking “Hey, we’re running a bit of a race with other companies that are trying to get themselves going again” – is AI something they could look toward to help them get there?
Deborah Leff (11:40):
So I think that’s going to be really critical. I think there is a couple of phases to what we’re going through right now. So right now it feels like Rome is burning and our focus at this moment in time is how do we be a good global citizen and use IBM’s resources to be as helpful to the global community as possible. And there’s a number of things that IBM is doing. We are part of a consortium with the White House that is making compute power available at a massive scale to accelerate research and discovery for doctors, scientists and anyone who is working treatment and cures. We’re working on all sorts of areas of resiliency and helping companies with the new normal, like the sudden work at home, and we’re working on being a trusted source of information.
Deborah Leff (12:40):
We’ve launched a whole site with together with weather.com.You have to go to weather.com/coronavirus and you’ll see county-level case information and trusted information about treatment and information. In addition, we are trying to be there as much as we can for our clients. So for example, one of the things that we had done for trusted information is pre-train a Watson virtual assistant for people with questions around COVID-19, like where to go for testing. I mean, you can imagine that influx of questions that all government agencies have been seeing, like a deluge of calls of people desperately looking for information. Our customers have experienced that as well. One of the very first places we made this available outside of government agencies and the CDC was a children’s hospital that had seen a huge spike of calls into the nursing emergency hotline to the tune of like 5 to 10,000 a day.
Deborah Leff (13:50):
I mean, how on earth can you have enough resources at a time of crisis to be able to handle those things? So we made our Watson virtual assistant available with the pre-trained models for COVID-19 and in 72 hours that was fully implemented and fielding calls from concerned parents. And then we started thinking, wow, you know, we have commercial customers that are also suffering in a time where they not only have to deal with the crisis around them, but they have employees with questions, they have customers with questions. And so we’ve opened that up as well for them to use the system for a few months at no charge, no commitment, just to help them through this period of time. So for the first phase, is this Rome burning? Like, what do we have to do right now to handle this crisis, to keep our business going,
Deborah Leff (14:43):
to take care of our employees, which I know so many companies are really struggling to make sure that they do best thing they can by their employees and to of course take care of their customers. I see firsthand how our grocery retailers are fighting so hard to make sure that there are the right goods to keep our families, you know, everyone going. We’re going to make sure babies have formula and diapers and the essentials and I love our heros on the frontlines that are making sure that people that need the things in this country still have them, including all of the healthcare workers that put themselves at tremendous risk to take care of the population. But once this starts tapering off, I think we’re going to enter into a new phase and that is why now I don’t know that people can really think about recovery.
Deborah Leff (15:37):
I think that’s unfortunately on the horizon with so much chaos right now and these numbers still climbing. I think that hopefully within a few weeks we’ll start to see that the number of new cases starting to level off and start tapering and then I think it’s going to be time to be thinking about what, how do we return to the operational, like the levels of operations that we had. And they may not be the same level. There may be, you know, there may be things we need to think about differently. I know that there’s a lot of, when you talk about becoming a data driven organization, we sometimes see resistance in the marketplace because there’s so much experience and expertise and tribal knowledge. I see data sciences and engineering teams create models that evaluate data and have amazing insights and I’ve seen line of business leaders look at the insights and say, I’ve been doing this for 15 to 20 years. I know better- that feels like a black box or you know, just magical thinking. I’m going to rely on my gut. But here we are in a point in history where there is no experience, there is no one alive that has lived through what we’re living through now, where you see stores closed down, hotels across South Florida shut down, airline routes back to what they were in the early 1950s. Who has the experience to make the critical decisions about how we bring that back online and how do we make the right decisions to accelerate the recovery? And I have to imagine that there are going to be things that people can be doing. I know there are going to be things that people can and should be doing. That data is the only source that’s going to be able to inform those decisions. And I think the reliance on AI and machine learning is going to be so critical that we will look back and we’ll say
Deborah Leff (17:50):
how and why did companies get to AI at scale? Because they had, they had to recover and that was the only way that they could know they were doing it as efficiently as possible while protecting their solvency. Yeah. That’s fascinating. You bring it up because it makes you think about things you have in the past that if you strip away all of the experience, what you have left in a lot of cases, all you have left is data or things that could be collected that you can somehow assimilate and make some sense of. So it makes me think a little bit about all of the data that you can put together. The way in which you synthesize that data a lot of times lead to conclusions which can lead to either foreseen or unforeseen consequences. I’m very interested in what IBM’s doing in terms of Watson being able to answer the phone, being able to answer questions because you can see how economic anxiety or health anxiety leads the right people react to certain things. Are there ways in which AI can take that data and somehow discern what’s the best message that goes out to people just by looking at data or is it always going to require some kind of human experience to add to that so that the data isn’t interpreted in ways that cause unforeseen circumstances to occur?
Deborah Leff (19:15):
So I think there are really two sides of what you just asked. So I think from an influx of questions that we’re seeing from employees and from customers as well. Well, I think for those types of things, a natural language capability gives you is the ability to train it with what those answers should be and then answer a very wide range of questions. So, you know, we have seen, you know, people have implemented over time chat bots that are really just automating FAQ if you will versus using a conversational AI assistant that allows the company to very quickly train on, you know, what is the intent behind the question so that they can answer that question appropriately no matter how the question comes in. That was the work that IBM has been doing as a service to the CDC and to citizens to create those models which are already created so that we can just make them available at no charge to help government agencies.
Deborah Leff (20:24):
and the CDC through that crisis and updating what the answer is is simple. It’s almost as easy as going into an Excel file and overwriting texts. That’s what it would respond. All of that works in any channel. So whether it’s a phone call or whether it’s someone asking a question in Slack or Facebook messenger or you know, through text, it doesn’t make a difference all coming from the same central place. And we have folks that we’re talking to, they’re like, listen, we implemented a chat bot a while ago. Yes, it’s FAQ’s. Yes, it took us a long time to train, but we’re getting so many phone calls that we need a way that we can quickly put, you know, conversational AI behind a voice bot. No problem. We got you. We can handle that. So I think for those types of questions and answers, you want to program what the company wants the answers to be
Deborah Leff (21:18):
and the conversational AI bot minimizes the training. It’s hosted online so that they don’t have to tax any resources worrying about running the environment to make it as easy as simple as possible. That doesn’t have to be a longterm solution but the stop gap that’s perfectly acceptable and I think that’s something that I would say is definitely trained on the other aspect of that is where we don’t have any experience and expertise to help us navigate the recovery. I can tell you that we are right now working internally within IBM on a think tank where we are thinking through what are going to be the right leading indicators. If you’re retailer and your stores are closed (and it’s going to happen) then we all know there will be a point in time where this crisis will be in the rear view mirror. It will be something we all lived through and thrived through
Deborah Leff (22:18):
because I believe in the resiliency of our human race. There will be leading indicators that will help inform those decisions. So whether they are things like the county level information on new cases or if I was a retailer I might be looking at the patterns of illness and recovery. The other thing that I think about is we were talking earlier about how people have prioritized certain decisions over time. I’ll just give you one example. In the world of retail, if a company had a brick and mortar store that might be driving 85% of their revenue versus 15% on their digital property, well, you know, something of that nature, they might have prioritized decisions to make investments and optimized things that impact the physical store and not necessarily have optimized the digital presence. So if you look at the way, you know, search engines on sites, recommendation engines, you know, personalization, the way you connect an individual with goods, all of that are going to be areas of investment. Because if digital is all you have, you need to optimize digital to make sure that you can capture every ounce of revenue that you can to aid the recovery. So we we have an internal think tank where we’re putting together all of these ideas so that when companies are ready to start talking about what’s the most efficient way to emerge from this crisis, then we’ll be ready with a lot of very mature ideas and even some pre-trained models that we can help them implement.
Kenny Mobley (24:08):
Yeah, that’s a very interesting point you made about what’s going to be happening as people and businesses get ready to move forward. I’ve read some things that say, maybe contrary to things we may think, that the recovery itself will probably take phases or stages over some time. It won’t be that suddenly all the stores are open back up and we’ll just start on a Saturday showing back up. But there may be certain segments in the U.S. that start working or certain industries. Do you think, and this may not be a direct AI question, but a technology question, that there are things being worked on now that will help in that recovery to have an understanding of maybe how we reopen the United States, for lack of a better term, once this is starting to get behind us?
Deborah Leff (24:56):
Oh, I absolutely do. I think that’s exactly what I was referring to and the fact that we don’t have the, you know, well the last time we did a global shutdown of all of our properties shows how we brought our system back online and this order and this is what was most effective. None of us have done that. So there are going to be brand new decisions that people are going to have to make and there are absolutely leading indicators that we’ll be able to harvest data likely outside of an organization that are going to become critical input to make the best decision making possible.The level of creativity that companies bring to the table is actually going to be their competitive advantage. I think that we’re going to be writing about this time period for a very long time, this is going to end up in curriculum in terms of, you know, what happened, how companies deal dealt with it, those that survived and those that thrived.
Deborah Leff (26:01):
If you look back in any recession or depression, there were always companies that thrived in that time period. I do believe that AI and data and the insights from proxy data is going to be exceptionally meaningful. This is what’s going to accelerate getting to AI at scale faster because it became necessary. Those that really went after thoughtfully trying to put the best collective minds together to strategize on what data to use as proxy data and what real time data of things that we can collect and incorporate that into the decision. I think it’s going to have a huge, hugely important, absolutely dramatic impact.
Kenny Mobley (26:52):
I love that way of thinking about it. Debra, you know, when Andrew Yang was running for president, one of his big platforms was how automation and AI may hurt and it very well may hurt at some point in some future. But I think the idea is you’re bringing forward now is yeah, there’s all these very powerful capabilities of AI, but you have to mix that with creativity and decision making and the real human element of it as well in order to really make the solution what it is. It’s a holistic thing, not a partial thing where AI just makes a bunch of decisions for us and we wait for the outputs. I really liked that. You think creativity may be the main driver between the really successful, the marginally successful and those that don’t make it at all?
Deborah Leff (27:41):
Oh, I think we can train the machines to do a lot of things, but we’re never going to be able to train them to have, you know, that level of ideation and that creativity you have, per say, with a data scientist. Just watch how iterative the processes and how it’s about finding different threads and following those threads to see where they lead us. It’s not so black and white. It’s that dedication to wanting to keep exploring and experimenting from there until you get to the point where you feel very confident that you can havemeaningful insights from it. That’s why we’re working on this now because by the time that our customers have found their feelings in this new normal, we want to be ahead of that curve so that we can provide even more value to them.
Kenny Mobley (28:36):
I would suspect that IBM’s doing a lot of work now on the data collection side so that if something like this does happen again, and we should presume it will, that it won’t be a first time again. Right. A lot of stuff and new information and data and ways of handling it have been uncovered, discovered and captured. What’s happening here? I’m guessing that IBM’s collecting tons of data right now as part of their collaborations with the government and the weather channel and other organizations. Is there a way in which that data is going to be used in the future? Are there new initiatives that are starting about learning about this as an academic pursuit so that it can be used again in the future?
Deborah Leff (29:20):
Absolutely. In fact, I was on a call a call earlier today with a member of my team, Bill Higgins, and he was a key driver in a lot of the work that we’ve done to drive the county level data into the weather channel app and on their website. And I have to tell you that this is going to be a fascinating case study because the way that it is going, I mean, you can imagine how much data it’s ingesting, structured, unstructured, how we’re scraping for it. But bringing all of that together in a trusted way is something that if we had sat down in a conference room, you know, five months ago before this was ever even a remote reality, you would have looked at this and said, okay, you want to do what? Wow, this is probably a six or eight, eight month endeavor and feels really big and overwhelming and should we even be doing this? The work was done, you know, in a matter of days by very dedicated people who worked literally around the clock. What I love about this story of what they’ve accomplished was, okay, maybe you’re not going to do these things in days. But even so, it now proved out that it’s doable in a short period of time. And I really hope tha they will do a deep dive and a postmortem on that project cause it’s enormous in scale and it’s amazing how quickly resources mobilized to just unite and get it done. That work is going to live on in some type of crisis response system. There’s things that we’re learning through this and how unprepared we all feel for what we’re going through.
Deborah Leff (31:09):
Now of course tere’s so much work that we’re doing on security and resiliency and disaster planning that was not in a lot of people’s disaster recovery. You know, if a system went down or if a server went down or if the data center went down, but nothing to this level and I guess we could be more prepared for it. But more than that, I’ll tell you that what I see on a smaller scale, I see that as very easy and not just in business but in our lives that we get into a business as usual routine. And sometimes when we’re in the business as usual, there are things that could be better. There are things we know aren’t working as well as they should.
Deborah Leff (32:00):
But we also feel like we lived with them all of this time. How urgent is it? We fix it. There are supply chains that have been operating with siloed, rigid, disconnected systems and they knew that they couldn’t get real time information on demand. They knew they couldn’t appropriately predict what their inventory levels should be but they could also continue running the business. And it was, it was, it was good enough. I think this crisis exposes is many companies, the, the underlying systems that needed modernization that we hadn’t quite gotten there. And now it’s very important that we do. And we will look back and not just say that this is the time that we got to just AI at scale, but it will have driven so much digital transformation because it really exposed the things in the organization that absolutely positively had to be fixed so that we were ready for the future.
Deborah Leff (33:00):
The one thing about September 11th is that it was so sudden and so impactful and then there was a quick return to normal as fast as we could. It was like, you know, the flicking of the light switch – went on and went off. It went off. It went back on the other way. Whereas here it was something we saw happening in a far off land. We watched it as it then impacted Europe and then it came to the United States. How did we not expect it was going to come here? It caught so many of us by surprise and then the speed of it swiftly moving through the country and the responsible things to do in terms of the way that things, everyone had to shelter in place and shut things down and to minimize social contact. Those are all the right things. But no systems were prepared. No commercial systems were really prepared for this.
Kenny Mobley (34:03):
Had they been, maybe some of the decisions wouldn’t have been as extreme as they are. Like you say shelter in place and others say the safest, best thing to do is just avoid contact. But given no other options you kind of have to go with the harshest of all possible responses and maybe learning more in the future things could have been done differently and that’ll be interesting to find out later as we’ve had time to look this over again, look back over all the data and have a good understanding about it. I agree with you. There will be books, case studies, articles and things in schools about this for many years to come. Looking back at this and how it was handled and how we should go forward and companies like IBM I think will be very prevalent in those discussions because of the way they have been able to help because of the resources at their disposal, the minds at their disposal and the creativity inherent in the organization and I think that’s great and I really appreciate your time today speaking with us about all that. How should people connect with you if they want to know more about all the stuff we’ve been talking about today?
Deborah Leff (35:12):
Oh, absolutely. The best place to connect with me is on LinkedIn. We’re publishing different things across IBM and I do my best to try and keep my newsfeed updated. Certainly anyone that would like to connect and reach out directly, I would say that, you know, IBM has really focused on just being as helpful we can and making sure that we all come out of this on the other side as strong as we can be and have a bright and healthy future together.
Kenny Mobley (35:49):
Yep. And that’s, of course what we all hope for is everybody’s health first and then minimizing the other things that go around it. So thanks again. We appreciate your time and look forward to talking to you again on another topic.
Deborah Leff (36:03):
Thank you so much.