What Does AI for Social Good Mean?

An interview with Rosana de Oliveira Gomes, who is a Lead Machine Learning Engineer at Omdena. Omdena is a global platform between mission-driven organizations with AI engineers, data scientists, and experts from diverse backgrounds to solve real-world problems. It was great to hear more about her current work, as well as exploring a perhaps unusual career move from astrophysics into AI for social good.

Read the full transcript below and listen to the full podcast here.

Topics explored include:

What Does AI for Social Impact Mean for You?
The Urgent Social Challenges Faced by Humanity That AI Can Help to Tackle
The Skills Which Can Be Transferred From Academia Into the Field of AI
The Key Challenges When Practising AI for Social Good
The Impact on AI for Social Good From COVID19
Advice for Someone Looking to Change Their Career Path
How Enterprises and Companies Can Best Engage in AI for Social Good

Nikita RE•WORK - Interviewer [0:47]

So I want to just jump straight into a topic that we'll be exploring quite a bit in this discussion. So Rosana, can you describe what AI for social impact means to you?

Rosana [1:01]

So thank you, Nikita. And thanks for the opportunity to have this discussion here, and also for a great initiative from your work for bringing women in AI to discuss so many different topics. So just starting answering your question, the thing is that AI is all over the world, we know there's this big hype about AI, right. And of course, everything we do, whether it's finance, marketing, business, or something more like medical, this all has an impact on society. But I think that usually what this umbrella of AI for good or AI for social impact means to have a positive impact in the world, and the outcomes of AI that can help people.

Nikita RE•WORK [1:53]

And what do you think are some of the most urgent challenges faced by humanity that AI can try at least to help tackle?

Rosana [2:03]

So I think the two main challenges are; so right now we are in the middle of a pandemic, and we have all the challenges from medical and healthcare, and AI can be really powerful to help to just speed up the process. And the second one, which is probably the first one that we have in the century, is the climate crisis, this is the reason why I chose to change careers. And within the climate crisis, you have lots of collateral damage that is caused by it so we can use AI, for example, for renewable energies to just reduce our CO2 emissions. But it can also be used for all these collateral damages, for example, you have extreme weather events such as cyclones or hurricanes, and they can help with disaster response operations. And also, with agriculture, when you have droughts and floods, they're supposed to be happening more often in the century. So these are all applications and there are many others, but just to mention a few.

Nikita RE•WORK [3:16]

And that kind of links on to one of the topics I mentioned at the beginning. So you have quite a varied background, can you share a bit more about your story on your transition from astrophysics into the AI for social good space?

Rosana [3:31]

This is a rather long story, but a very nice one. And so actually, since I was a child I wanted to be an astronomer. I found out that the sun was dying, but it's like in billions of years, and I was like, oh my god, I need to do something about it. And so I decided to set up in astronomy, and this is what I did for many, many years. So I've been in academia for more than 10 years, and I was researching astrophysics. And during my Ph.D., I came to Germany. So I came to do an internship at the Frankfurt Institute of Advanced Studies in Frankfurt, and this is an interdisciplinary Institute. This was the first time that I was not only interacting with people from physics, but also with people with several other backgrounds, people from neuroscience, people from math, and then they were all applying their knowledge to other problems in the world. And this kind of opened my mind to realize how many other things are out there and how many of those skills that I have can contribute to the world. And this was also in 2014 when I went to this institute, and back then in 2015, we had the Paris agreement for reducing CO2 emissions and tried to help fight climate change. And I think all these things together kind of started to resonate with me and I started to understand what is possible to do as a scientist, and even as someone with a technical background because those things sometimes are seen maybe as more like humanitarian things or a humanities problems. And to see that, as a scientist, you can actually contribute a lot to society as well is quite interesting.

Rosana [5:23]

So this is the story. Then I also had my mentor. So my supervisor, my Ph.D. supervisor, here in Germany, was also working in astrophysics, but he was working with renewable energies and other fields as well as deep learning. So being in this institute actually, kind of made me realize the broader applications of my knowledge, and thanks to my boss, I also could start working with deep learning and seeing other possibilities. Unfortunately, he passed away last year, this was something that was very unexpected and this was the moment that I realized that I can either continue researching in academia or I can actually go outside of academia and try to do something more applicable to the world. And so that's basically how I decided, at the end of last year when all the students that were still in the group ended their masters. I was helping them with all the situations and so on. So I was the postdoc there in the group, and then I was helping them. And then when they all finished, I thought, okay, so now I'm going to take a break and take, kind of like a sabbatical if you want, to find out what is out there for me to do in the world. And this is how I came across Omdena actually. And since March this year, I've been working with them. I've been working already on several topics, both in the field of AI, but also in the field of AI for good, such as disaster response, and online abuse of children, and sustainability, and so on. So it's been quite a beautiful journey so far.

Nikita RE•WORK [7:23]

It certainly is such an interesting background. And something that comes up quite a lot on our podcast, as well as at our actual events is exploring that link and route between academia into more real-world applications, I guess. So from your experience, what skills can be transferred from academia into the field of AI?

Rosana [7:50]

So of course, when you talk about academia, you have all sorts of fields out there. But specifically, I can talk about my field. So I have a background in theoretical physics, which was applied to astrophysics. So whoever has this scientific background already has a lot of knowledge regarding data science, even when they don't realize it, because if you're in a lab, even if you're doing simulations, or if we're doing something more experimental you will be dealing with data. So everything you do with data analysis, if you want, you're a professional analyst, for many years. But actually, what was very surprising to me was to realize that I was already expecting that all the knowledge that I had in terms of math and statistics and so on, would be very helpful in programming as well. But realizing that, the soft skills that you have to learn, so if you are in academia, and you are already used to communicating at your work in conferences, you are used to being in projects, international projects, as well. And if you are more senior, you already have a lot of experience with leadership and with, for example, mentoring students or you know, carrying out your own independent research. And I think these skills, apart from critical thinking, you just having this problem-solving approach, these are more important than technical knowledge because you can just take an online course and learn it fast.

Nikita RE•WORK [9:36]

I think that's applicable to lots and lots of situations. So touching back on the AI for good topic, what are the key challenges that you've come across when practicing AI for social good?

Rosana [9:54]

When we talk about AI for social good, of course, this can be done at the level of a startup. They have a kind of a society oriented goals such as renewable energies or mobility or something like this. It could also have this in big tech companies with technical applications and so on. But you have a lot of AI for good happening in NGOs. And in this circle, I think there's still a lot of challenges. I think it's mostly related to this very traditional institution. So the times you're talking about institutions that exist for hundreds of years, and they've been doing things in one way, and then there's also this entire hype about AI. So I think sometimes people look at AI in a little bit suspicious way of what can be done. And sometimes you even have to use different words such as technology or innovation, to just not use the word AI because people get a bit, you know, taken aback, and also have lots of problems such as, for example, they don't have the data for analyzing or they have the data, but this is still, for example, the things are not digitalized. So having to deal with this kind of problem, like I've been talking a lot to people from NGOs, that are data scientists in NGOs because this is actually a few of my interests. And usually, the problem is that you have this kind of very difficult way of dealing with bureaucracies. So this is still something to change. And sometimes, it's also not clear what AI can do for this kind of institution. So they have those problems and then you have this problem of communication between the data scientists and the people that are doing the military operations and to just put these people together and communicate, and so they understand what their needs are and what needs to be done.

Nikita RE•WORK [12:00]

Exactly. And it's important that we do have that mix of different stakeholders involved. So touching on that, how can enterprises and larger companies, that have perhaps more resources, best engage in the AI for social good space?

Rosana [12:18]

So I think that there are many ways of engaging social groups such as, for example, you see lots of companies nowadays talking about having more sustainable products and so on. And this is very good for the world for sure, but this is not necessarily related to AI. But I think when it comes to AI, one thing that is very, very serious that we need to address is bias. So we have nowadays, pretty much every company in the world is this trying to have some AI team, some data science team. And we have to be very mindful of bias. Because, you know, if you are, for example, a company specializing in finance, and you're making risk analysis for credit, and this is not taking into consideration minorities, women, people of colour and so on this becomes a serious problem. And then it might come to a point where we cannot change this anymore. So I think this is something to be very mindful of.

Nikita RE•WORK [13:19]

Definitely. And looking a bit more, I guess, at current times, how do you think, if COVID-19 has had an impact on AI for good? And has it advanced that in any way? Or what are your thoughts on that?

Rosana [13:35]

So yeah, I think that whenever there's a crisis, you have those typical sentences that there are opportunities. And I mean, there are of course, like negative things. So one thing that I've noticed is that because I'm also transitioning into careers, and so on, it's as I mentioned before, from the NGO sector, there's still this kind of, they are still figuring out what AI can do for them. And if you have something risky, something that they don't understand, the investments that will be done in this field can be kind of limited. So this is kind of where you want a negative aspect. But on the other side, what we see a lot nowadays is a world that, at least in my perspective, is way more connected than before. So we are in lockdown in some parts of the world, we are practicing social distancing, but on the internet, we all have meetings all the time. We are always connecting through work and especially like when you think about online schooling and work, these things had to be completely digitalized very fast. And this is something that AI can help. This is something that, for example, when you think about AI for cybersecurity, even now exactly right now I'm working on a project that is about online abuse of children, so if they think that we have to put all those children online, and they need to be interacting with everybody in this environment, this has to be safe. And this is something that AI can assure help with. And this is something that I would be expecting to grow more.

Nikita RE•WORK [15:20]

Definitely. And in these ever-changing times, what advice would you give to somebody, and perhaps one of our listeners that are looking to change their career at the moment, as I'm sure there will be quite a few people out there doing so?

Rosana [15:38]

This is actually a question that I get a lot. So one of the things that I do in my free time is to give advice to people teaching careers. So especially, like, you know, we always have people contacting me via LinkedIn, and so on. And I always try to help students and women and you know, minorities. And this is a question that I get quite often. I think that one thing to remember when you start and when you decide to change careers is to realize that you're not starting from scratch, there are always transferable skills that you already have, something that you already learned before, they can be used in a maybe a little bit of a different way, but in different contexts, if you want, but you already know. So you're not really like I have zero experience of this. The other two things that I usually say are, you have to find out, what is it that you want to do? What is your passion? Because changing careers can be quite frustrating. You know, in some sense, you will be a junior, again, and so on, this is not a very nice experience. And then you kind of have to have this motivation, right. And if you're doing something that you're just changing careers, just because they don't have this drive, it's difficult. And I think that when you think of the field that you want to work in, it is important to understand which skills are necessary for this field because, for example, let's say you want to change careers to AI for medicine, in this field, you probably need to know more about computer vision. And whereas if you go to finance, you need to learn something else. So to kind of direct whatever you need to know, and to see what you already have and what you still need to learn.

Nikita RE•WORK [17:34]

And what are your thoughts on the challenges faced by women in AI practitioners in particular?

Rosana [17:43]

I haven't been in the field of AI for such a long time. But I've been a physicist for basically my whole life and so the problem that you see, or the challenge that you see is that the women are simply not there, you know, like, you don't really see women in those spaces there are for AI or computer science, and so on. So I even had a look and we had around 22% of AI professional women in the world. And this is really not much. And this is because, if you go to, you know, now and again we have webinars and events, conferences, if everything you see is just a lot of white men, it makes it very difficult to convince young girls that they can be part of this. And if you are an older woman, it also means that you have to work twice as much to get the same recognition. So we have a problem there and we need to find ways to motivate women to continue in this field. So we need to create spaces to show that we can do as well as everybody else and I think this is very important.

Nikita RE•WORK [19:07]

Definitely. And role models and mentorship come up in every conversation that we have on this podcast. And rightly so it's hugely important to have that visibility for the next generation coming into the space and also for people that are looking to start their career in this space later in life as well. So thank you so much, Rosana, for sharing your bit about your background and your journey and your career into AI. It's really inspirational to hear that you had another successful career before working in AI and other interesting projects that you're doing now as well. So for any of our listeners that would like to get in touch with you or to find out a bit more about your work. What's the best way to do that? Is it Twitter or LinkedIn or what do you think's the best way for them to contact you?

Rosana [20:00]

Thanks for having me, it's a very interesting conversation. I think the best way to contact me is via LinkedIn and I also have my email there. So if you just want to reach out and say hi, I usually reply in a couple of days at the most.

Nikita RE•WORK [20:17]

Fantastic well we will share the details with that in the notes below on the podcast page. But thank you again, it's been fantastic to speak to you. And hopefully, we'll catch you again in a few months to hear a bit more about where your work is led to.

Rosana [20:31]

Thank you.

Nikita RE•WORK [20:33]

Thank you, thanks very much.

Nikita RE•WORK [20:40]

A huge thank you again to Rosana for taking the time out of your busy schedule to chat with us this week. I really enjoyed learning more about her journey from academia to applying AI to solve real-world challenges. If you'd like to have the opportunity to meet more inspiring women working within AI, then be sure to check out our upcoming virtual women AI evening, taking place next month on the 11th of November. We can't wait to bring the community together again, this time online. Until next time, take care.