Crazy Smart Asia: Stephanie Sy—How AI Will Save Or Destroy The World
Depending on your point of view, big data and AI will either save the world or be the end of us.
Stephanie Sy’s data science consultancy, Thinking Machines, builds machine learning models for organisations including the World Bank and Unicef, using data to tackle the biggest issues facing the planet today, including poverty alleviation, the climate crisis and, of course, the fight against Covid-19.
But while Sy's company is a shining example of how AI can be harnessed to make the world a better place, she’s far from blind of the perils of the misuse of data. In her conversation with Gen.T editor Lee Williamson, former Googler Sy covers everything from the privacy paradox to fake news, weighing in on the issue of regulation along the way.
A self-described “basement nerd”, Sy also talks about the struggles she’s faced shouldering the responsibility as the company’s public-facing CEO, why she’s never accepted VC money, and why, despite appearances, 2020 might be the best year to start a company.
ON AI SAVING THE WORLD
“AI doesn’t save the world on its own; it saves the world by making humans superpowered. Think of AI as an Iron Man suit. It enhances our capabilities, but the driving force behind it is still a human heart and mind.”
ON THE COMPARISON BETWEEN DATA AND OIL
“People say data is “the new oil”. Yeah, it is like oil. It's toxic, it's expensive to store, and it might just explode on you. I want CEOs and government policymakers to collect the bare minimum necessary. Defence is so much harder than offence. It’s so hard to maintain a secure database, and you can't be breached if you don't have the data in the first place”
ON FAKE NEWS
“Tech platforms are the new gatekeepers [of truth]. They just don't want to admit that they're gatekeepers. At least journalists knew that it was part of the professional code”
ON HARMFUL ALGORITHMS
"Social media companies always argued that they're a public utility. Facebook has always thought of itself more like a phone company than a news agency. But that metaphor doesn’t work, because on social media you aren't just talking to your friend—you’re getting exposed to new audiences via an algorithm.
"AI is hurtful because when it has just one objective, it will find strategies that are optimised towards that one objective. The YouTube recommendation algorithm, for example, wants people to watch as many seconds of video as possible. Without a human specifying it, over time the algorithm has figured out that what people want to see is more conspiracy-oriented, is more of the same, is more radicalising. There have been some very interesting studies on how the YouTube recommendations algorithm is a radicalising force, pushing people to conspiracy-oriented channels."
ON DATA ANONYMITY
“It’s not possible to anonymise data. Carnegie Mellon did a study in the year 2000 that showed that 87 percent of Americans can be uniquely identified with just their zip code, birthday and gender. I'm pretty doubtful that anybody can stay truly private and anonymous and interact in the digital economy.”
ON THE NEXT CIVIL RIGHTS MOVEMENT
“When I was in Silicon Valley in the 2005 era, everybody was talking about how they wanted to use technology to save the world—to the point where it's a parody these days. But that spirit still exists in Silicon Valley; it exists in tech companies. But now they have to make these very hard choices that are political in nature.
"They aren’t just part of the tech community now, they’re part of the civil rights movement. They have changed the world—now they have to figure out “Oh, great. What do we do now?” And I really believe that technologists around the world are all going to step up to the occasion and figure out what it means to be a technologist and a citizen.”
ON THE DOWNSIDE OF VC MONEY
“VC money is like jet fuel. You should take it if you want to go someplace very, very fast. But if you put it in your tiny Honda Civic, it’s going to explode. Taking VC funding is not inherently bad, but your choice of financing has to be totally aligned with who you are as a company and what your mission is. Thinking Machine’s vision is to get organisations to make good decisions by building good data systems. And that’s something that happens over a 20-, 30-, 40-year timeframe.
"The issues we’re dealing with, like climate change and poverty, they aren’t something you can solve in a seven-to-10-year VC cycle. So we looked for sources of funding that were more aligned with our time scale, which ended up being customer revenue. So that worked out really well for us [in 2020]”
ON WHY 2020 IS THE BEST YEAR TO START A COMPANY
“I think venture capital has to get smarter in this next year. But, weirdly enough, this is a good time to start a company. When everything is going extremely well, when funding is easy, it's hard to get noticed in the noise. Starting a company in a recession also forces discipline. You can't BS your way through anything. The value you deliver has to be real because nobody is going to give you revenue for a vanity project. And because the whole economy is doing badly, you can pick up really good talent who otherwise would be working at a tech company or a conglomerate
"So if you are in that category of company that's very disciplined, that creates revenue really quickly, and you learn really fast, you can grow. Maybe not at the fastest possible pace, but you grow in a way that builds product-market fit. Then when the recession lifts, you're in the perfect position to go from zero to 100 really fast.”
Listen to all released episodes and subscribe using your preferred podcast platform on the Crazy Smart Asia hub page.