Data Driven Chat: “We didn’t do a very good job of building resiliency into really important industries.”
Ganna Pogrebna in conversation with Briana Brownell — What COVID-19 showed us about data science, how life after COVID-19 will be different, resilient systems and recommended reading and watching on AI.
Academic, Educator, Consultant, and Blogger Ganna Pogrebna started Data Driven Chat to explore the many interesting aspects of behavioural data science including human behaviour, data science and AI. Ganna Pogrebna is Professor of Behavioral Analytics and Data Science and a Lead of Behavioral Data Science strand at the Alan Turing Institute — the national centre for AI and Data Science in London (UK). The Data Driven Chat is the official podcast of the Behavioral Data Science Special Interest Group at Turing.
Data scientist turned tech entrepreneur Briana Brownell is the founder and CEO of Pure Strategy Inc. Using the leading edge technology in natural language processing, machine learning and neural networks Pure Strategy’s Automated Neural Intelligence Engine (ANIE) gives organizations the insights needed to make decisions confidently.
The conversation originally appeared on Data Driven Chat. Click here to listen.
This is Part 3 of a three-part series. You can read Part 1 and Part 2 here.
Ganna: We cannot avoid the topic of coronavirus and you mentioned the amazing work that you are doing in the medical industry for med folks, so I just want to ask you a little bit more about the role of data science in solving the current crisis.
There is a lot of speculation, like some people who say that, ‘Oh data scientists should only get involved when there are field experts working with them!’ and others say, ‘You know we all should be doing work on COVID and just see what happens!’ so on which side of the spectrum are you at? What do you think about this?
Briana: I think it’s really interesting because a lot of the really successful applications of data science and AI in fighting coronavirus have all been human augmentation systems or human-centred systems that are being helped along by good data science and good AI.
An example of that would be some of the triage applications. So, being able to analyse a chest x-ray, for example, and predict whether that person is at risk for complications or for serious effects.
Instead of having doctors or other health care professionals like radiologists look at all of the images and do it in a slower way, you have a system that can automatically surface the cases mostly likely to have complications so the human can analyse those images first. You augment and make the job of the physician and the radiologist and the health care professionals that are working at the hospital easier and empower them to provide the best care that they can and empower them to deliver the best outcomes that they can. I think is really interesting and important.
“You empower health care professionals to provide the best care that they can.”
Emerging patterns has been really interesting in terms of the coronavirus too. When AI systems originally saw the blip of ‘Hey this might be something,’ way back in January or December, all of the early warning systems that surfaced it went to a group of human experts who then dug deeper into it to see what the real risk was. What was interesting about that is the humans who were looking into this were experts, professionals in infectious disease and they recognized that there was a real potential for big impact.
I love the idea of data science and AI systems being used as human augmentation systems. It’s not ‘AI is going to save the day!’ and ‘Oh we have this tech system and we’re going to replace you.’ Instead, it’s ‘How can we use AI and data science to make humans better, make us more effective, and empower us to have better health outcomes.’ I think that COVID-19 has really proven that AI and analytics is not going to replace people. It’s going to augment them.
Ganna: What challenges do you think are in intersection between decision science and data science today? Are there any new challenges that we saw with covid or other problems that were there and that were all of a sudden highlighted by the current situation?
Briana: The biggest problem that we found was the brittleness of all of our systems. If you look at all of the huge industry-wide system failures that happen due to covid-19 they’re all because of that brittleness of the systems. You have hospitals with no surge capacity or no equipment because hospitals are run to operate the most efficiently as they possibly can and so why have extra? Extra doesn’t make you money. It makes you lose money, right?
“If you look at all of the huge industry-wide system failures that happen due to covid-19 they’re all because of that brittleness of the systems.”
Supply chain systems — When all of the restaurants closed and people started shopping in a completely different way, you had huge shortages in things like yeast. Like, I don’t think I can still buy yeast at the grocery store because consumer behaviour was changing. The buyers were changing. All of a sudden, restaurants just weren’t buying any more food. People’s habits on what they were eating completely changed.
I think that it really exposed how brittle some of our optimization can be to tiny little changes. I remember reading that there was one farmer who was producing onions for restaurants and large-scale production. He wasn’t able to sell it directly to consumers because they were being put in these enormous bags and they didn’t have a system in place to be able to put it in a bag in the size that a household would be able to use.
It’s really exposed how we did a great job of optimizing our supply chains and making everything as efficient and perfect as possible but one little movement and all of a sudden those systems completely break down.
What COVID-19 has really taught us about how we build our systems is that we didn’t do a very good job of building resiliency into a lot of the really important things, the really important industries we need to. Like supply chain for food, of course, is one, healthcare is another. Education systems, how all of a sudden teachers and parents are scrambling to try to teach their children when they’re at home and everybody is trying to work from home and all of these IT systems need to be set up to allow workers to work from home.
There are all of these systems that we’ve created that are super brittle to any kind of change and so I’m hoping that in the future resiliency is going to be seen as a good thing to build in into these systems.
Ganna: So how about this future? Where do you see it, apart from building resiliency in supply chain, do you see any other important hot topics for data science? And the intersection between decision science and data science and the future?
Briana: I think that the changes in consumer behaviour, the changes in the way that we live our lives, this has been a permanent change. We shouldn’t be looking back to “when are we going to get back to normal” because it’s going to be a new normal. It’s going to be the next normal. And it’s going to be very different from the world that we lived in pre-covid.
Things like working from home are going to become very common. There are already huge companies that are saying that they’re going to have a permanent work from home policy. That’s going to have a huge impact on rural areas or places outside major centres because all of a sudden, if you can work from anywhere that has a great broadband connection, then you can stimulate the economy with jobs even if there is not a physical office there.
“I’m encouraging everybody, if you have data about your organization, pre-covid, you really have to check that your assumptions are still valid. Any decision that you made off of that data you need to really think about whether it still applies.”
Travel, especially air travel, is going to be fundamentally different. I think that we are going to stop having large airports. Large airports are going to basically become a thing of the past. We’re going to have multiple small airports because, imagine how much safer you would feel if a small airport was completely sanitized and you weren’t at risk with hundreds of thousands of passengers in a large airport. So the results of COVID-19 are here to stay and we’re going to have a rocky next couple years where we adjust to the next normal.
Data and analytics is going to be an important part of that because we’re going to be able to see what the changes really are and which changes are becoming permanent. I’m encouraging everybody, if you have data about your organization, pre-covid, you really have to check that your assumptions are still valid. Any decision that you made off of that data you need to really think about whether it still applies.
Ganna: I really like your point about how data propagates through the entire supply chain systems. I think many people did not realize that how much analysing today’s demand actually has impact on the decisions about supply and how this volatility is. The normal circumstances are fine, but if something like this happens, then your business can completely collapse because you’re not prepared to understand that the shocks in demand. You just are not very resilient, like I said.
So we’re almost at the end of the interview and I just have one last question. It’s a traditional question: If I asked you to recommend one book and one film, what would you choose? What would be your recommendations?
Briana: The best book about artificial intelligence in my opinion is “Machines Who Think” by Pamela McCorduck. This book was written several decades ago, I think 40 years ago or something, when AI was starting to generate more and more interest. It is absolutely one of the most fascinating books about the topic because she marries some of the humanities and some of the other fields like art, literature to the topic of AI and looks at it in a much more holistic way than any other technical AI book. So that is a hundred percent one you should check out. She also has a biography that just came out that I absolutely loved. I went through it really quickly— read it start to finish — even though it was really long because she’s had an amazing life. So I would say that one is absolutely number one on my list.
Instead of a film can I pick a TV show instead?
Ganna: Yes it’s a popular choice. A lot of people do shows.
Briana: So I’m going to go with Battlestar Galactica. I love BSG, the new reboot, and the reason that I love it is because it brings up so many ethical questions about how we’re going to deal with sentient AI in the future: What the risks are. What kinds of systems we might have to put in place. And it’s just an absolutely fascinating show.
Ganna: Well thank you so much Briana for finding the time. I know you’re very busy, but thank you for this. Good luck with everything and continue all the amazing work that you are doing. Definitely we’ll keep an eye on what you are up to and Pure Strategy Incorporated what you guys are up to!
Briana: You’re very welcome, thank you.
This is Part 3 of a three-part series. Click here to listen to the whole conversation.
Part 1: Data Driven Chat: “It was something that really mattered.”
Ganna Pogrebna in conversation with Briana Brownell — The chapters of her data science journey, impactful projects and current work in health care
Part 2: Data Driven Chat: “The potential of data is not realized in most organizations”
Ganna Pogrebna in conversation with Briana Brownell — Underrated skills in data science, getting executive buy-in, and how we can all shape the future
