Back in September, when I was doing some research, I contacted Clive Thompson. He’s a columnist for »Wired« and author of the book »Coders« (here’s what the »New York Times« and »Süddeutsche Zeitung« said about it) in which he describes the growing influence of »a new tribe« and how it’s »remaking the world«, as he put it.
After talking to quite a few of the mainly young, white and male programmers he’s quite critical of their work. They are good at solving logical problems and breaking big challenges into small ones, he writes. But too often they only understand little about the complexity of the world, politics and how people other than themselves live. Which is important to keep in mind because »the civic role of code is only going to grow«. It will not just exist in Twitter, Facebook or search engines but also in future cars, TVs and fridges who plan to supposedly fulfill our needs by collecting our data.
The story I was working on had a different approach. I was wondering how the evolving collaborations between Silicon Valley Big Tech on the one side and social start-ups and NGOs on the other side are playing out. Both have started working together in numerous climate and sustainable projects and I wanted to take a look beyond the technical aspects and talk to people who could answer some of my questions. Clive seemed to be one of the them and I was glad that he accepted my interview request.
The story was supported by the Heinrich Boell Foundation’s »Transatlantic Media Fellowship« and has been published on their website (the German version was first published in the November issue of the tech magazine »t3n« and can be found here). But as there was only very little space for his answers in the text I thought it would be of interest to publish a longer version of our conversation (here’s another one with scientist Lynn Kaack). Clive agreed, so here’s my Q&A with him.
Clive, Microsoft, Intel, Amazon Web Services, Salesforce and Google have started »AI for Good« programs and claim they want to solve climate change, sustainable issues and social problems. Why are these tech companies putting effort into this?
I think it’s threefold. First, many people in these companies are getting genuinely concerned about the environment and they like the idea of being able to use corporate resources to try and help. Second, it’s that this can help with recruiting top engineering talent. Not all of them, but especially younger ones like the idea of working at a company where they know that some of the resources are going towards solving real world problems across the planet. The third reason is that they’re also trying to do something ostentatiously good to counteract any bad publicity.
It’s digital greenwashing then?
I don’t think it’s entirely that. At the top level for the CEO it’s indeed partly to clean their reputations by spending several hundred million dollars on social and climate projects. That’s nothing unusual. But some of the people working on the projects are super sincere. Google has several quite significant environmental projects. For example they used AI and partnered with nonprofits that do satellite monitoring of everything from groundwater to deforestation. They help building AI prediction models to pinpoint where deforestation is happening before humans can see it. They’ve also done an enormous amount of work with groundwater. One of the UN climate goals is to prevent groundwater loss. That’s incredibly crucial for farmers and citizens but the UN and many countries have very bad data on how much groundwater they have. And that’s important information in order to manage the remaining water. So, are companies like Google motivated by greenwashing at the top level? Yeah, sure. Are these also important and really good projects? Yes. Both things are true.
In 2020 Microsoft announced to set up a »planetary computer«. They want to join forces with environmentalists and scientific institutions and store all the relevant data from these partners in one cloud so that it’ll become easier to solve current and future problems.
I’m always deeply concerned about centralization of power, especially when it comes to trying to solve public issues. We’ve already seen the dangers that can happen when a lot of public information is collected at a place like Facebook, right? They know tons more about all of us and we know about each other and about them. That’s dangerous for democracy. I don’t know very much about this »planetary computer« so I can’t give you any specifics. But there are several questions to ask: How open is that data? Can other people use it and see it and know what’s in it? What about the privacy implications for individuals, countries and cities? And if this data turns out to be incredibly useful to management – is Microsoft the only one that controls it?
As you mentioned some companies are working with NGOs and social start-ups. These organizations often rely on the help and grants they’re receiving because they don’t have the technical skills and talents. But how serious are Microsoft, Google and AWS about their collaborations? Is there a risk of them dumping these unprofitable projects once a new CEO takes over or budgets are being cut?
Is it possible programs could vanish and the company loses interest? Absolutely. This is the danger of governments not having these capabilities themselves. The only way to truly sustainably keep a program afloat that has wide benefits for everyone is to have a good government program. It is capable of saying we are going to spend a few billion dollars on something that will only have long-term gain instead of just short term gain. That’s not at all the calculation that companies make. They exist to make money and quickly pivot away from things that don’t make money. But when you speak of AI in particular I can’t really name any cutting edge governments projects in the United States. I don’t know if it’s different in Germany but if you want to pursue this work it’s widely understood that you do that in the private sector.
It’s the same here in Germany, but it has not always been like that in the US.
In the 1960s and 70s when they were doing space exploration the single best place to learn the really cool stuff was NASA, was with the government. People actively said I don’t care if I can make more money working for IBM, I want to work for NASA because I get to do this with the smartest people in the country in a room. And that’s just not the case with AI.
»What are the other areas in face recognition we’re not looking into yet but should be? It’s probably a much worse problem than any of us can imagine«
Is there any way corporations can turn climate or sustainable goals into a stable business making profits?
I don’t have any specifics but look at solar energy. It’s a classic example of an industry that required significant subsidies globally for decades to get to the point until it could produce efficient panels. And if we were to drill down into it, we could certainly figure out all sorts of business opportunities including ones that if governments were thinking more strategically they could intervene in. Think about weather collection. For the daily weather information in the United States, Canada and in the UK they created a National Weather Service to gather data from around the country. And they started doing that in the early 20th century because they realized they were a farming nation and Farmers needed really good information about what the weather is going to be. And they were not getting really good information because you need someone who’s going to collect the data across the entire country everyday and analyze it.
No company could have profited off placing all the sensors and temperature stations all over the country.
That was way beyond the scope of any for-profit corporation in the early 20th century. So the government did it. But then they did something very smart. They made all that data completely open. You don’t have to pay for that weather data in the public domain and this produced billions of dollars of commercial activity. Governments in the west are not doing this big bold thinking very much lately.
In your book you make clear that coding is a monoculture. It’s mainly white, young, males who have problems with human interaction. Should we be concerned that they are working on real world problems such as climate change?
Having spoken to techies involved in these projects I think most of them have a broader view than your typical young coder. Most of these folks are a little older and have some experience with multiple fields. You need to have that to be interested in climate change in the first place. What I’m more worried about is the narrow way of white guys looking at things. They can often seize on things that they think are very exciting because they can imagine turning them rapidly into a business. And they promise the moon but there’s really not much there. I’ve seen so many silly little attempts and apps to save energy that never really worked and vanished after a year or so. They present not so much danger to the environment and more to social and media issues where you really can create a company at your laptop, go up there and completely break the world. When it comes to environment, very, very few young techies are trying to do the big hard things.
When it comes to AI a much discussed problem is biases in algorithms. Does this play out when it comes to using major data sets in climate or environmental issues?
That’s pretty much always a problem with major data sets. I’ll give you an example. Let’s say you were going put out subsidized solar panels based on collected data around where people seem likely to actually install them. The machine learning identifies 30 neighborhoods that would be perfect for our penals. The problem here is the historical data on the sales. It’s going to be heavily weighted towards well-off families that can afford the stuff because in the early days it’s always expensive and so the only people who install panels are the well-offs. Whereas if you were to make it right you would gather data from a heat map of the city which shows you the urban canopy of trees. This data clearly reflects the fact that poor neighborhoods are a lot hotter. They have less trees and more asphalt. Historically over the years mortgage companies would not lend to those neighborhoods. So people own their houses less which leads to governments and developers build a freeway right to the middle of it and put in shopping malls instead of planting trees.
So if I started working on such a business I would have to be careful about the data that guides my decisions?
If you don’t watch out you’re going to have all sorts of crazy bias in there. There’s no way around that. The smart and active people now understand that and work really hard to gather and think about the data before they process it in their AI. Unfortunately there’s many more lazy people than there are active people.
There’s a least one NGO called Thorn using face recognition software which AWS has temporarily banned from selling to the US police because too many problems occured. Should social start-ups and NGOs continue using it?
In general there have been massive problems and many inaccurancies with face recognition. It’s been a hot point here in the US. I know about the technology’s use by police enforcement and the danger the privacy problems present to the public. Everyone sells it with any bloody camera for no reason now. But I don’t know about other areas and I feel like that’s a great question. I wish I knew the answer. What are the other areas we’re not looking into yet but should be? It’s probably a much worse problem than any of us can imagine.