Gillian Tett is the author of Anthro-Vision and is featured on the Book Club Podcast

In 2005, Gillian Tett wrote her colleagues at the Financial Times a memo arguing that the paper should devote more time to covering those submerged parts of the ‘Financial Iceberg’, such as credit and derivatives.

“The biggest risks in the world are not usually hidden through any dastardly, James Bond-style plot,” the FT’s US editor-at-large, who predicted the 2007-8 financial crisis, says today.

“The bankers weren’t concocting a wild scheme to bury what they were doing with financial innovation in 2005-6 into some kind of dark tunnel. The cultural patterns meant that most of the problems were hidden, but not seen.

In her new book Anthro-VisionTett explains how her PhD in anthropology helped to predict the financial crash. This was because she realized that siloed bankers were blind to the wider context of the products being sold.

Tett talks to the World Economic Forum about COVID-19 and the climate crisis as well as artificial intelligence. Book Club Podcast how taking a worm’s-eye, empathetic view of risks facing the world today can help to explain – and potentially solve – them.

Gillian TettCultural assumptions that we learn from our environment shape the human condition. Culture doesn’t exist as boxes, it’s a spectrum of difference. One of the most important things we need to do in a world that’s both globalized and polarized is recognize that the cultural assumptions we each inherit are very powerful, but they’re different. It’s a good idea to try to understand the lives and thoughts of others. However, it also helps us to see ourselves more clearly.

There’s a wonderful proverb which says: ‘A fish can’t see water’. We can’t see our own cultural assumptions and biases, unless we jump out of our fish bowl, go and ask other fish what they think about us, and then look back at ourselves. When we look at that wider cultural context, we begin to understand why using mere quantitative tools to navigate the world, like corporate balance sheets or economic models, or big data sets, simply isn’t enough to capture the complexity of our cultural experiences today.

Gillian: COVID-19 has made it very clear you can’t beat a pandemic just with medical science, or big data or computer science. You need to combine it with social science to understand the cultures and behavioural patterns that shape how people do or don’t behave. And that extends right across all the problems which we’re facing today, such as climate change, income inequality, or both the promise and peril of big data and artificial intelligence. My plea in the book is to start blending these different disciplines.

There is good news: evidence is available to support this. After several months of trying to beat Ebola with medical science, the World Health Organization (and others) made a shift to embrace behavioural science in their policy during the Ebola outbreak in West Africa. That was the key to ending the Ebola outbreak. A similar process of learning – sadly a slow one – has been happening with COVID-19. There is a greater appreciation for the necessity to combine these disciplines when it comes to vaccination roll-out. These same principles should be applied to AI, tech battles, and climate change. There is some progress, but it’s still very patchy.

Gillian: Firstly, we have to understand people’s cultural experiences of climate change issues. In a conversation between [actor]Robert Downey Jr., the US climate envoy on the World War Zero website. Downey Jr. said that most people are afraid to talk about climate change and feel guilty. We have to find a way to communicate these messages really effectively to consumers, and be sensitive to what is or is not blocking action by them, just as we’re having to work out why some people today are not taking vaccinations. We can’t assume that what makes sense to an American or European policymaker will make sense to consumers in other parts of the world.

The second is to look beyond the edges of your model. For so many years, economists treated the environment as something which was external to their economic models about how the economy was going to behave, so-called ‘externalities’. The environment was a footnote in corporate finances and business people. They believed all these resources were free and could be ignored. Corporate tools like economic models and corporate balance sheets are really useful – but they’re always defined by the limits of what you put into the models, they’re bounded. We have to learn how to see beyond these limits and to understand the context. The environment is a part of the context. This is forcing us to rethink corporate accounting and economics.

The third way is simple: it’s a human tendency to put others in a box and shun people who seem different, but anthropology argues we can’t do that. We’re all interlinked in a spectrum of cultural difference, in a chain of humanity, and when the weakest link of that chain breaks, we often all suffer. We saw that in COVID, we saw the perils of ignoring what was happening in faraway lands, or pretending you didn’t really know or care what was happening in Wuhan. The same thing is happening with climate change. We cannot afford to ignore other people who seem different from us in a world that’s so tightly integrated as a global system.

Gillian: Having a diversity of perspectives in any workplace is really, really important: it might take more time to get to an answer, but it’s less likely to be really stupid. One of the reasons the 2008 financial crisis happened was there simply weren’t enough checks and balances in the financial system. The majority of people who worked in West financial companies had the same mindset, training, intellect, and perspective. They were often all men and plagued by groupthink and tunnel vision. A common view is essential to help business leaders who are often lacking it.

Anthropology believes that it is worth trying to immerse ourselves in the minds and experiences of others in order to see yourself differently. It helps us to think about all the things you ignore in your everyday world, the so-called social silences, the parts of our environment that we tend to overlook because they’re so familiar, or because we’ve labelled them as boring or geeky. Social silences are never irrelevant, they’re often crucially important for explaining how the world really works, and how we reproduce the patterns we have around them for the future. It is important to look at social silences from other perspectives.

Gillian: Yes, in some narrow ways: there won’t be another crisis caused by subprime mortgages again and I predict there probably won’t be a crisis caused by a shortage of capital in the regulated banking system again. That is very encouraging. But I think there’s still a need to be more imaginative about forward-looking risks, and recognize that threats almost always crop up where there are silos, they’re almost always found where problems fall between the cracks of existing institutions. It is easy to forget the importance of cultural patterns and incentives, tribal behaviour, and social dynamics.

Gillian:In the tech sector, we have seen a similar pattern of groupthink, intense tribalism, and tunnel vision in recent years. That’s worrying. In the lead up to 2020, there was a tendency for people to ignore social science around medicine. A column I wrote some years ago said that people were overlooking what was happening in the geekier corners and risks of pandemics. It was similar to the way they were neglecting the geekier corners before 2007. Today we ignore some of the major ethical questions raised by AI and finance.

We’re frankly also ignoring many of the issues around AI in general today, because once again, technical knowledge is held in the hands of a tiny group of elite technocrats, who the rest of the world tends to ignore because their activities are labelled as boring and geeky and dull, and therefore not of interest to everybody else.

Gillian:It is crucial to understand both the context in which AI is created and used, as well as the tribalism and tribalism of the coders writing AI programmes. We’ve seen the way that you get embedded biases, by the lack of diversity in the coding teams. We must also consider how AI is implemented in product development and whether those who implement it can understand the ethical and social context. Alex Karp, chief of Palantir (one of the largest tech companies in Silicon Valley),, pointed out during the last year’s IPO filings that we’re putting enormous amounts of power in the hands of a small group of computing elite, who operate in silos, and don’t necessarily want that level of power and probably aren’t equipped to deal with it.

AI is a powerful tool. It basically works by collecting vast amounts of data points about our human activities, looking for correlations, then extrapolating into a future. But there are limitations to this. If you’re looking for data points about what we do and say, you tend to ignore social silences because what we don’t say by definition doesn’t get recorded. Secondly, correlation is not causation: you can’t understand why people are doing things if you assume that everything can be judged just by looking at correlations and data points. Thirdly, context change, which means that what happened in the recent past doesn’t always reflect what’s going to happen in the future.

Culture is not a clear-cut, linear pattern that can easily be analyzed using Newtonian physics. It’s an incredibly multi-layered, contradictory, baffling entity that exists in a spectrum of difference, not boxes, and is constantly changing in subtle ways. It’s more like a river.

AI platforms can scan financial markets and look at medical data. They can also design rocket ships and play Go. However, no AI platform has ever created a good joke. Jokes are used to define or articulate social groups. You must be part of a group to share cultural assumptions in order to get a joke. Jokes are able to play off all layers of culture, including social silences. AI programmes are useful but have limitations. Therefore, AI needs Anthropology Intelligence to be most effective in today’s world.