Humanity defines the challenges of artificial intelligence

Podcast Transcript 

Ross Dawson: Hello and welcome. I'm Ross Dawson, a futurist, speaker, and author, hosting The Where The World's Moving podcast series, presented by OFX, a global money transfer specialist for Where The World Is Moving.

Ross Dawson: Through this podcast series, we explore our fast moving, ever-changing world, and some of the extraordinary technological and social shifts that are shaping business and society, driving collaboration and globalization, and supporting human progress. To find out more about the exciting ideas in this podcast and the rest of the OFX series, go to WhereTheWorldsMoving.com.

Ross Dawson: The field of artificial intelligence or AI was established in 1956. Yet for many decades it failed to fulfill its promise, however, in just the last few years new approaches have brought extraordinary advances. The amazing capabilities of these machines are now making us redefine the relationship between technology and humanity. Ultimately, the potential is to bring out the best four human. I can think of no one better to speak about this than VP of AI at Sage, and a leading AI ethicist, Kriti Sharma. Welcome to the show, Kriti.

Kriti Sharma: Hey Ross. Thanks for having me.

Ross Dawson: Artificial intelligence, it's supposed to be intelligent we presume, with the word intelligence in there, but it is artificial, as in we've made it by humans. So, how humanlike is AI? Is it as intelligent as humans or not?

Kriti Sharma: Ross, before we go into the debate about how intelligent or humanlike it is, it's even more important to understand what's the value of it, and whether it should be humanlike. For example, the AI community itself is very divided between machines that are pretending to be human or trying to look very human or mimic human interactions versus those that are explicitly able to explain themselves as, "Well, you're now talking to a machine and I'm here to help you."

Kriti Sharma: We in popular culture, connects to all of it. We have a certain images of robots stomping over the world and stealing all our jobs and getting involved in our private lives and developing deep, human connections with us. But in reality, this technology is already everywhere in our lives. For example, when you log onto your social networks, on your Facebook, the feed is optimized by AI. When you're searching on Google and it predicts the next few words, that's AI in action.

Kriti Sharma: Or, if you use Siri, Alexa, or type something on your phone and it predicts the next few words, or autocorrect that sometimes goes wrong. It's very much about the applications of AI in our lives and designing them in a way that it's adding value to humanity rather than creating the uncanny valley scenarios where we're creating creepy machines that look like humans and of course, the data and privacy issues that come with it.

Ross Dawson: What are some of the ways in which AI can add value to us?

Kriti Sharma: I'll give you examples. The most prominent ones at the moment are in for instance, healthcare, where if you are living in a community or an area where it's very difficult to get access to a healthcare specialist or to see a doctor, an AI could do preliminary diagnosis. In London just yesterday, there was a study published where an AI could with 100% accuracy, detect issues that needed urgent care for eye related issues, eye diseases.

Kriti Sharma: Now, this is tremendously powerful for healthcare professionals who would otherwise be under-resourced and can't reach everyone who needs urgent care at all times. AI is not replacing the human doctor but it's augmenting their capabilities by prioritizing cases. The right people get attention at the right time.

Kriti Sharma: We also see examples that seem a bit more science fiction but are closer to reality than we thought. For example, self-driving cars, where I was just in New Delhi last week and I was stuck in traffic for two hours, for something that would take me 30 minutes. Now, if I had a self-driving car, that would be pretty awesome, 'cause I could be reading the whole time, or just watching videos or whatever else I would like to do and improving the productivity in human lives.

Kriti Sharma: It also has tremendous value towards accessibility. For instance, people who are visually impaired or can't drive for instance, AI could be a great tool or aid. I also want to call out its value in our workplace, in our business life. We are already seeing coworkers who may not be human. For instance, just at Sage, we a couple of years built a tool, an AI chatbot, assistant, smart assistant, that works with you and does all the boring and mundane parts of your work life.

Kriti Sharma: For example, churning out reports, completing and filing for invoices, and chasing for finances, HR stuff. Things that you would outsource to a machine and the machine would happily do it, so the humans can focus on the more creative element.

Ross Dawson: All of these supporting humans in what they already do by prioritizing or being accessible or doing the things which we don't want to do.

Kriti Sharma: Exactly. And also identifying patterns that humans can't see. Because AI has the potential or has the ability to sift through millions or billions of transactions and data to detect anomalies. For example, fraud detection, or cybersecurity related challenges. In fact, I see AI as a tool also to fix some of the issues that technology has created. For example, if you look at in the online sphere, misinformation, AI could be a great tool to detect what's true information and what's not. Same with online bullying, cyber bullying scenarios.

Ross Dawson: These are human things, these are not the things of machines, and these are things that are laden with value, so can AI have values. If so, how can we teach the values to what is an emerging intelligence?

Kriti Sharma: Absolutely. It's very important to think about when we're designing or building AI systems, including many of those that are in our everyday lives, it's very important we teach them the right ethical values and principles to perform and survive in human society. You can do it in various ways. You can do it at a design level, and also from data and opinions perspective.

Kriti Sharma: A lot of the AI that we see in our lives every day, it's in the form of Siri, Alexa, Google Assistant, that we speak with every day. Now, you might notice that many of them, in fact most of them have very feminine personalities and female voices. They're turning your lights on and off and ordering your shopping, or playing your favorite music.

Kriti Sharma: You get tools like IBM Watson or Salesforce Einstein, and they're making important business decisions for you. To me, that's a fundamental issue. We are even stereotyping the machines that we're going to be interacting as a gender stereotype that we're seeing. We've got to teach machines not to necessarily be like humans or be human intelligence levels, but to reflect a better society, something that's more diverse, and so that we're not repeating our mistakes from the past.

Ross Dawson: How do we do that? If we have I suppose, this focus on diversity and balanced values, how can we as human designers of AI bring about those kind of perspective or balance in the AI?

Kriti Sharma: First up, it's very important to have teams that are as diverse as the society. Now, many of your listeners might already know that there's a huge gender issue, gender balance problem in the tech industry. Only about 15-17% of the tech workforce is filled with women, and that's an issue. It reflects in a lot of consumer products we see, the extreme stereotypical behavior that we are already seeing in technology.

Kriti Sharma: There is also a racial diversity issue. For instance, quite recently MIT Media Lab, Joy did a study there, a great researcher in AI, and she found that a lot of the commercially available facial recognition, facial detection algorithms, including those that are available by major providers like Microsoft and IBM are extremely racially biased. They have much higher error rates for darker skinned women than for white men.

Kriti Sharma: That's an issue, 'cause this technology is coming into our lives, when you're trying to log into your computer, or log into your phone, you're going to login processes to access a banking app, or in some cases, even surveillance where nation states are using that information. Now, it's very important to create technology that's diverse and it works for everyone. In this case, it couldn't even detect Oprah Winfrey and Michelle Obama. Just because of the lack of diversity in the understanding of AI.

Kriti Sharma: We need to make sure that there are diverse teams creating AI and we need to make sure that AI is given diverse data to learn from. The issues that we see are of inequality and bias in technology, are not necessarily coming from malicious or a bad place. It's just because we haven't fed the AI a diverse dataset to learn from.

Ross Dawson: What are the things that we can do to be able to make sure that we have diverse teams, that have the experience and expertise to work together to create that?

Kriti Sharma: Firstly, we've got to recognize that the most important technology of our times, the responsibility to create the most important technology of our time, should not be left to geeks like myself. We need people with very diverse skillsets. AI is as you rightly pointed out in the beginning, it has been through an AI winter where the technology had great promises but it took a long time to start adding value to the real world.

Kriti Sharma: At the same time, it's becoming more accessible today. That's a really positive. However, it's about intelligence and intelligence isn't just about teaching a machine to perform a task or for it to become very efficient at solving a problem. It's also about creativity, emotional intelligence, empathy, and problem solving, decision making. These are all skills that machines are not very good at, but humans are.

Kriti Sharma: It's about finding that balance and building that relationship. In fact, AI is getting so clever that it can even replace the job of software engineers and it can write its own code. So, if I do my job well, I might not have a job anymore. What that means is it's not just the role of the techies that's important in the world of AI, but it's people who understand civil society, anthropologists, creative thinkers, writers, artists, and they need to be the right participants in giving these skills to the AI machines that they don't have today.

Kriti Sharma: Another point to remember is unfortunately we, those who are part of the AI community, there is a sense of elitism or inaccessibility where there's this mindset or view that you've got to be a super smart rocket scientist to work in this field. Which is absolutely not true. I'd say if we did a survey just over a year ago, about young people and their interest in tech, and whilst we still had about 20-25% who were very excited about building a career in this technology, we asked the others why not, and it was very interesting what we heard.

Kriti Sharma: First reason was that they didn't think they were smart enough, which is a self-limiting belief, and secondly, they were not getting the right skills in schools. Some of the kids were smarter than their teachers in computer science classes, and just better at solving the problems which creates a different dynamic. 'Cause these kids are, they're born with these digital technologies. They've never seen the pre-internet world, and lastly, they thought they would rather be a YouTube star or a podcast host.

Kriti Sharma: I do believe that there is an opportunity to bust some of these myths that you don't have to be super smart, rocket scientist, with mathematical skills to contribute to AI. One of the roles in my team, for example, is a conversation designer. It's someone who writes the personality of AI that speaks to how should it learn to communicate with humans. Now, you can't learn that in computer science courses.

Kriti Sharma: I think now we're moving towards a time where it's not just about teaching everyone to write code, but it's going to be about embracing the human element. Giving humans the time to be more human and one of the my frustrations with the tech world, and why I focus so much on ethics, is because the promise of digital technologies was to make us more productive. But we've ended up becoming more addicted to this technology and we're spending a lot more time on our devices.

Kriti Sharma: I believe that AI, if we design it ethically, we have a huge opportunity to make our society, make our workplace more productive, rather than what you're seeing with a lot of other digital tools, where we're spending more time online increasingly than offline.

Ross Dawson: Do you see that possibility of how these technologies could help us be more human ourselves?

Kriti Sharma: Absolutely. I think so. Even in the business context, when I talk to a lot of companies at Sage and our clients, we often get, for example customer service, that's a very interesting area. Where we used to have those IVR systems where it doesn't matter what numbers you type, you're never able to speak to a human. But now we're creating AI technologies that an AI can solve about 80-90% of the issues, and queries. But when you do need to speak to a human, it can be smart enough to detect when it's the right time to hand off. When you do speak to a human, the conversation is very high value.

Ross Dawson: Does this mean that we need to develop new skills? Are jobs redefined because of this?

Kriti Sharma: Yes, I absolutely think so. I also think we need to work more collaboratively between people with different skillsets. For instance, the tech industry needs to open up to have people who are experts in the social sciences and the humanities, anthropology, psychology, the legal frameworks, and the creative people like yourselves, you should play a big role in building and designing this technology for the future.

Kriti Sharma: For the young people, we need to make sure that they understand that developing their creative and problem solving skills, and they know how to use AI and consume AI and developing great applications. Just to give you an example, we are running a program in the UK called Future Makers Lab where we are training young people who are not the typical kids who go to code clubs every Saturday, and love all things tech.

Kriti Sharma: These are kids who might not have much interest in tech. We teach them the basic skills of AI like how to create a machine that can do voice recognition, that can speak back to you, that can understand what you're saying, and is able to see things, computer vision. Then, we leave them to work on any project they want and come back and present. 9 out of 10 times, they come back with a project that has very strong social purpose.

Kriti Sharma: For example, one of the kids is building a tool for refugees, to translate signs into their own languages when they're trying to integrate into society or a community. Another kid is 13 years old, he's using AI to detect and tackle the fake news or misinformation problem. Another girl is building an AI to help the elderly who live alone. None of them yet, have tried to build a killer robot that will destroy the world.

Ross Dawson: Good.

Kriti Sharma: It gives me a lot of hope. My mission is to bring down these barriers between people who can create the technology and everyone else, and I think we all have a huge role to play. It needs to be a bigger collaboration and participation between different parts and elements of society.

Ross Dawson: A lot of potential in terms of what this technology is giving these children, for example. We've moved into a society where so many people are stuck to their screens and they're interacting with technologies, less useful and more useful ways. But are there ways in which AI or related technologies can help us connect to other humans, to help build community through collaborating or do you see that kind of potential?

Kriti Sharma: AI could help us be more productive by prioritizing what needs to be done for us in our digital lives. We could even cut down our on screen time by up to even half, 'cause technology can tell us, "These are the things that need your attention." You don't need to check your WhatsApp notification, and then look at your Gmail, and then look at your Facebook notifications, and then look at your work emails all together. AI can help us develop good habits and prioritize our screen time.

Kriti Sharma: It can also recommend you to do more physical activity online, and we're seeing a lot of that with fitness trackers, for instance. Your connected IoT devices or smart devices that are trying to do that in a very interesting way. I would say we need a lot more of this design thinking, about human productivity in the process of creating AI than we do today.

Kriti Sharma: Also, we need to start thinking about teaching machines the right ethics, the right values from the beginning, from day zero, rather than creating a technology and then trying to retrospectively fit it into a certain framework. When I teach the kids, I start teaching them about how AI can learn different behaviors and how it's important to teach it the right values from the beginning. When they learn AI at the same time, they learn how to do it ethically. That could be an interesting experiment to try with grownups, too.

Ross Dawson: If you're enjoying this episode, listen to the rest of the OFX series at WhereTheWorldsMoving.com. 

Ross Dawson: Some of the potential of AI where we can personalize our interactions with businesses, we can personalize healthcare, but we can also personalize education so that education is relevant to the individual. How do you see this playing out in terms of how AI could help us build better education?

Kriti Sharma: Yeah. I'll start by sharing my own example. I studied in Rajasthan in India where in my high school we had 70 students and one teacher. That's 7-0. It's not uncommon in large parts of the developing world to have this kind of education system and scenario. What this meant was the teachers were only able to focus on the top five and the bottom 10.

Kriti Sharma: This was one of the reasons I got so fascinated by AI is when I was a kid, I thought, "How awesome would it be if I had my own personal robo-teacher and it could work at my own pace and help me learn at a speed that was comfortable for me?" My ways of learning were very different for each subject and as I grew and developed, and it was one of the motivations for me to get into this industry, to think about how we could give every child their own personal tutor and I'm just really excited that we are very close to it today than we were ever before.

Kriti Sharma: In fact, one of the kids in our Sage Future Makers Lab program is working on an AI tutor to help kids revise at their own pace or various subjects and topics. That's a small example that we could do. If we do this at scale, we give everyone the opportunity to learn at their own pace. Particularly for kids with learning disabilities and learning issues, where they can do that at their own pace.

Kriti Sharma: Also, those who speak various languages in very diverse parts of the world. I was in Australia just two months ago and I saw the huge amounts of diversity within the community and the society there. If an AI tutor could work with kids from different backgrounds, those who have different mother tongues in various languages, first languages that they speak, that'd be pretty awesome.

Kriti Sharma: More connectivity, more smarter interactions, 'cause just if you look at how AI's working today, there's a lot of personalization that needs to happen for each one of us, but at the moment there's still a lot of it is done in groups, or for example, all the young people 15-25, they think like this. A lot more cohorts. We could get a lot more personalized information, more connectivity and smart devices around us, and at the same time, I also believe it will improve privacy of our data and cybersecurity platforms that are being created. 'Cause the same technology could be consumed by those services as well.

Kriti Sharma: On the human side, I also think some of the recent events about data privacy in the news that we've all seen have led us to a greater awareness of what the value of our data is, and what does consent in the online world mean? How is our information being used? We are more aware than ever before, and that's a positive thing that has come out of it, and I believe in the future of this technology, there will be a lot more focus on how is information being collected? How is it being used? If it's being used for personalization, when is it good to use it? When is it not good to use it? And make conscious decisions to not do certain things, if they're not working for the benefit of entire community and society.

Ross Dawson: Yeah, well that's another aspect of education is learning, learning about that value of our own data. You mentioned this personalization and we talked about that in personalized education. What are some of the other applications of personalization that you see particularly valuable in healthcare or in shopping, or in our homes, or what other aspects where you would see this personalization that AI can bring us being useful to us?

Kriti Sharma: Yeah, so in addition to the consumer applications we talked about, in our homes and on our online lives and searches and social media and messaging applications, and healthcare, in addition to those, I think there's huge potential for AI to add value to businesses and improve productivity of nations. When I think about small businesses, it's usually one person doing all sorts of jobs and startups, and we're living in a world where there's a lot more freelance workers, more gig economy gigsters, as they're called, and more small businesses.

Kriti Sharma: There's one person doing several different tasks and chores, and I think AI could be a great tool to give personal, a digital AI CFO, or an AI HR advisor to each of those businesses. Luxuries that were previously reserved to just large corporations or companies, and this could boost productivity massively, 'cause now people are, businesses are doing more useful things, and of course what it means to the economic development of society.

Kriti Sharma: I think that's where it's going to be really powerful, and also because the cost of getting access to AI technologies and techniques is going down rapidly every year, and that means it's more affordable and within reach of businesses to consume.

Ross Dawson: There's a number of things you've mentioned which are essentially about this democratization, the giving to the many of this power, where one person can have all the resources, do their own business, or everybody has access to the education. In the big frame, what are the things that we need to be doing so that this is a force for everybody, that it is giving value to everybody and not just the wealthy or the large businesses?

Kriti Sharma: Yeah. We need to do a few things. There's a strong role for policymakers to play here, and that would involve in setting aside budgets and implementing large scale reskilling programs. Even though it's increasingly more available and easier for people to consume, we still need to do strategic policy level investments, in making sure that existing workforce that might be impacted, or they might have parts of their roles change because of AI, are ready for it.

Kriti Sharma: To make sure young people who are in schools today are not just taught how to code, but they're taught how to teach machines to do tasks, or train AI systems, and value all the human skills at the same time. There's a big role for the skills agenda and policymakers to do work in that field. Second is establishing an ethical framework for digital technologies including AI. This will include what's acceptable and what's not acceptable.

Kriti Sharma: For instance, if an algorithm is discriminating against a segment of society, something that a human wouldn't be allowed to do, then it should be equally enforceable for algorithms and AI use as well. For example, if you're using an AI to find a candidate to match with a job, which it is one of the most popular and well established use cases of AI in human resources, is recruitment. Now, if an AI is learning from historical patterns, is looking for the next CEO of a big pattern, it's quite likely to think based on the trends in society today that a male candidate is more likely to be a successful CEO than a female candidate.

Kriti Sharma: In many cases, there are examples where it's starting to filter out certain candidates on the basis of their gender, or their background. If a human hiring manager did that, we would be outraged, but algorithms can go undetected, because we are not looking into these issues with the right level of attention. Therefore, it is a strong role for policymakers, lawmakers, governments, to make sure that algorithms are learning the right ethical framework and data ethics come into the same category.

Kriti Sharma: Where if you're collecting information, you're consuming information, it needs to be implemented and used in the right way. Secondly, lawmakers, policymakers, need to make sure that AI is developed in a diverse and ethical way. We need to make sure that that accountability and transparency and diversity is being built at a much broader policy and a corporate framework.

Ross Dawson: In some ways, AI is helping us be more human but also it is forcing us to think through the ethical issues. Now, this is almost an opportunity or a wake up to be able to have the conversations in society about what we believe is right. What are our values? What are the ethics? Since we can shape society literally with AI, we can actually start to force these conversations, bring them to light, and to help them shape what we are doing as the human race.

Kriti Sharma: Absolutely. And now is the time. We are seeing tremendous applications in all sorts of worlds. Healthcare, education, transportation, public services, cities, smart cities, including on our phones and in our homes. Now is the time to make sure that we're designing this technology in the right way from the beginning.

Ross Dawson: You have a wonderful title and a wonderful job, and you're supporting all kinds of initiatives in your work through your expertise in AI. So, is this a career you would recommend to others? If so, what is it that they need to do to be able to work in this field?

Kriti Sharma: I would, but I might be biased. I think it's awesome to work in AI. I'll tell you about the first machine I built. It was a robot that would fetch chocolate Snickers from the snack bar.
Ross Dawson: Very good application.

Kriti Sharma: Solved a very big world problem for myself. I do better things with my life now, but it was really cool, 'cause I was still very young at the time and hacked this machine and it got smarter every day and it was really exciting to see that evolution of this technology. It's great fun to build AI techniques and they sometimes surprise, the algorithms surprise you by discovering things that you didn't know existed.

Kriti Sharma: Most importantly, when it adds value to people's lives, when actually gives them access to something they did not have before, it's really fulfilling. In terms of what to do to get into this exciting career, first thing, you don't have to necessarily go down the coding path that I did. If you're really interested in technology, get yourself, start learning some machine learning capabilities, mathematics, coding, however, that's not the only route to getting into AI.

Kriti Sharma: If you have any special creative skills, think about how AI could augment those and collaborate with someone. And most importantly, keep building. Keep building AI applications. It's much easier, very cheap to do that. There are several online courses on how to apply AI to create new applications. Just get started.

Ross Dawson: How is technology, not just AI, but other technology, helping us be more global, to cross boundaries, to be able to live in a more global world?

Kriti Sharma: Yeah. I'll tell you my personal example of languages. I'm not very good at learning new languages. I only speak two, English and Hindi. I refuse to learn any more, because I can see how machine translation is getting so much better. I did a lot of experiments between English and Hindi, and just with Google Translate, it used to be highly inaccurate, even five years ago.

Kriti Sharma: In the recent years, it's in the high '90s, in terms of accuracy rates, and it's only going to get better. I also think AI can help us create more sustainable communities and not only AI, but several other inventions that we're seeing in for example, material sciences and analytics, where if you look at the global and carbon footprint, we can optimize a lot of that with digital technologies.

Kriti Sharma: In the workplace, we are using a lot more of these techniques like video conferencing, telepresence, and more immersive techniques now also with augmented reality where we can create a more sustainable future for the planet by not needing to do air travel as much. That's just one example.

Ross Dawson: Absolutely. Your organization, Sage, helps many businesses around the world. What opportunities for business do you see enabled by these new technologies of connectivity and intelligence?

Kriti Sharma: I think being able to hire a diverse workforce and the nature of work that's changing for businesses, it has been enabled by technology. A lot more people who we're looking at trends and patterns where they're getting flexible working options, they don't necessarily have to commute in peak traffic times. The same applies to businesses, now they're able to create more diverse workforce, give them the flexibility. It's also improving the employee experience where they can access various services, they can use a lot of messaging and productivity tools at work. They can be connected as and when they want to.

Kriti Sharma: Those kind of tools are very powerful and they're reflective of the consumer life that we have. Business is getting very close to the great consumer applications we have of today. Giving them a lot of power to analyze the data and information and make informed choices is an area where we're seeing a lot of businesses use this technology. Instead of just guessing, we can give them very accurate information about which areas to grow in, how are they doing compared to other businesses of similar sizes, giving them that 100% of the time, 24/7 information and support that they need to keep doing the awesome things they're doing.

Ross Dawson: So to round out, many people were very aware of AI as something powerful that will impact the world, impact their lives, impact society, and somewhat overwhelmed by it. But what would you suggest for anybody, for all of us in how it is that we think about it? How do we approach AI and its power?

Kriti Sharma: I would recommend looking at problems that can solve, rather than the technology itself. You can even argue that tech is a bit over hyped at the moment, where every other company is doing AI. You must realize that AI is not necessary for everything. In many cases you just need good old automation, right? You don't need a robot turning your washing machine on and off. You just need a smart washing machine that can optimize the load and its environment, that sort of stuff.

Kriti Sharma: You don't really need to think of it in the form of what Hollywood wants us to believe, that there are humanoid robots is the only form of AI. We need to think about the problems it can solve, where we need automation versus just simple digital applications, versus intelligence with AI. We need to think about the problems that AI can solve for us in our everyday lives, and then make sure that we are working hard with diverse groups of people towards building them.

Ross Dawson: Well, it's inspiring to hear your personal journey, Kriti, from Rajasthan to where you are now. But also what you're doing to help others play part of that role. We're taking a human perspective on technology, I think you're a wonderful example, so thank you so much for your time and your insights today, Kriti.

Kriti Sharma: Thank you very much for having me, guys.

Ross Dawson: To find out more about how OFX is helping their customers move money around the world, go to OFX.com. This podcast is presented by OFX, and is a BBC StoryWorks commercial production.