«We think Big Techs want our clicks to flock us with ads. But what they are after is the huge amount of data that is necessary to build AI services». Self-driving cars, systems to manage energy consumption, testing devices to defeat illnesses… According to activist Evgeny Morozov data is like oil: they provide private companies with immense power over States. Reason for which, they should be a “common good”.
Talking to Evgeny Morozov is fascinating and terrifying at the same time. Because the Belarussian political scientist and activist, whom we met at the Feltrinelli Foundation in Milan, reads the small daily actions like pieces of a jigsaw puzzle. Things like putting a “like” on a post, registering on facebook in a dream city, unlocking a bike with an app, correcting a translation on Google … All this, says Morozov, is used to shape a possible future. Not a rosy.
«Those small actions train artificial intelligence systems (AI) that need human input to grow. When we will have provided enough data – 5, maybe 10 years from now at the most – we will no longer be needed. And the totally free services on the net will be a distant memory».
That, in itself, would already be a big problem: a recent MIT study calculated that if they were paid, the services that are currently free online would cost us between 15 and 18 thousand dollars each. But it is far from being the real scary side of Evgeny Morozov’s story.
«: mainly American ones (Google, Facebook, Amazon, Microsoft) but partially Chinee (AliBaba, JD, Tencent).
«Since data started to be used to train AI every online service has been designed to generate clicks»
Because they are the ones who have access to data that for almost 10 years have been using to refine Deep Learning. That is to train the machines through decisions made by humans. ».
People think that online services are free because companies earn with advertising. Instead, are they use us as a personal trainers for robots?
Evgeny Morozov: «Advertising makes money. But building a monopoly on AI services a lot more. The two goals have coexisted since 2009, since the ability of Deep Learning has been speeded up».
How does Deep Learning work?
«I explain it with an example. To translate a text with a machine, it has always been necessary to feed it with dictionaries and grammars of all languages. Then it was discovered that by exploiting enormous quantities of already translated texts it was possible – by turning words into numbers – to build computational models that represent the relations between words as they appear in correct sentences. So the computer does not need to know what the phrase means: it just has to make sure that the numerical representation of the completed period matches that in the other language. Which it can do instantly after digesting billions of sentences».
And now we have a more performing free translation service. What’s wrong with that?
«First of all it’s free now but in the future who knows. . Because Google used the texts written by the translators of the United Nations and the European Commission (financed by our taxes) to begin this process that we contribute to every time we correct the system. And remember that since data are used to train the AI each service is designed to generate more clicks. So how much do phenomena such as internet addiction (purposefully designed!) or fake news cost us? But maybe we should not worry, because eventually artificial intelligence will save us both from addiction and hoaxes. So they tell us those who created both».
So what should we do?
«It’s like asking what we should do about climate change. As individuals, faced with structural problems like these, I would say nothing. . We can therefore do something with our choices as citizens, not consumers».
Specifically, what you think could start the change?
«In the United States there is much talk of imposing a division of servers – the so-called “Breakup Google” movement. It is thought that what worked in the early twentieth century to block the monopoly of oil will work now with data. But when they are divided data lose a large part of their value. .
Instead of breaking up the tech giants, it’s better to force them to recognize that data does not belong to them, and force them to make them anonymous and accessible to the community.
Their use should then be regulated (for example, free for those designing public utilities, paid for those who have commercial purposes). But all this should be accompanied by a robust and proactive industrial strategy that allows those who have so far remained alone to watch (Europe) to exploit the talent that exists in abundance in its universities to build AI systems for the benefit of the community.».
Some cities, like Barcelona, are going in this direction…
«No city can compete with the computing power of Google, Facebook or even Uber, and probably not even a coalition of cities will. Also because their political and economic models are determined nationally or globally. This is why their ability to implement independent and effective policies is under constant attack. In the case of Barcelona, what truly helps is the creation of a “technological sovereignty”. That is, to enable citizens to have a say on how they operate and how they are to pursue technological infrastructures..
But many cities have gone completely wronge, believing in promises of greater efficiency guaranteed by start-ups, greater creativity generated by private hackletons and more transparency through open government initiatives. But many of these initiatives, rather than eradicating public corruption, provide reasons to cut sectors that actually work quite well. . That will be free as long as it is possible for them to collect and accumulate data».
If cities have a too limited power who could start this change?
«The states or, even better, the European Union, because what’s needed is courage, cohesion and firepower».
Should we be optimistic?
«It’s always nice to be. But frankly, how can you think that the countries of Europe agree on a theme like this one when they can not even find common ground in tax policy?»