Tuesday, 22 January 2019

What is that makes us "smart"?

Zat Rana
Let’s for a second imagine two very different people, with two very different backgrounds, studying the same thing in their own unique ways. In this case, it’s the sea. The first is a university professor, someone who is an expert on oceanography; the second is an old-fashioned fisherman.
The professor went out into the world, conquered its many challenges, eventually finding himself at the most prestigious of universities, learning at the edge of our collective knowledge. The fisherman, however, did what he was expected: He graduated from high school — itself was quite an achievement in his community — but then, he took over from his father, tending to the waters that surrounded them, just as his own father had taken over from his father before him.
Over the decades, these men studied exactly the same domain but from different vantage points, with slightly different purposes. The professor knew all of the forces governing the bodies of water on Earth, but he spent little time in the actual sea. The fisherman, of course, spent all of his time in the sea, but he knew little of the fancy terminology.
Now, let’s ask an interesting question: Who out of these men has a deeper understanding of how the sea works — the professor or the fisherman?
It’s tough question, and it’s also an ambiguous one. If your first urge is to ask your own question in response to clarify what is meant by “a deeper understanding,” I’d say that that’s a good step. Context here matters. And yet, when, in different forms, this question is asked in philosophy (rationalism vs. empiricism) or in psychology (Do IQ tests measure something meaningful as it relates to the lived world?) or in terms of the utility of logic (abstractions vs. reality), many people settle for one side and have a hard time reconciling the two in a way that does both of them justice.
At its core, this question is really a question of knowledge: How do we gain knowledge about the world? Rationalism says that it comes from our thoughts (from language, reason, and mathematics), whereas empiricism says that it comes from our senses (from observations, habit-patterns, and intuitions), and once this distinction has been made, each school carves its path further away from the other, leading to futile arguments that ignore the possibility that maybe simple reduction isn’t the best way forward here.
My own starting point is slightly different. First, I suggest that a better way to look at this is to distinguish between knowledge and wisdom, and then, I also suggest that we move away from the rationalist-empiricist dichotomy. In Buddhism, for example, there is no dichotomy because in many traditions, thought itself is considered to be a sense, just a more powerful one — in a way, a secondary one. Their starting point is consciousness, and from there, they see each of the capabilities of the human body — sight, sound, smell, touch, taste, and yes, thought — as a point of inquiry into the nature of reality.
It’s very clear that humans don’t experience all that consciousness has to offer. Snakes, for example, can see things in their field of consciousness that humans can’t. Similarly, dogs can smell things in their field of consciousness that humans can’t. This doesn’t mean that these sights and smells don’t exist in human environments; it just means that humans don’t have the evolutionary bodies that can tap into these different kinds of experiences. Hypothetically, if consciousness is an infinite dark field, then each sense can be thought of as a small bright light that illuminates one part of it to uncover reality. A dog or a snake’s field lights up different parts than that of a human’s field, but neither captures the whole thing.
The interesting thing about humans, of course, is that we have this faculty for complex thinking, which allows us to create knowledge. Now, what is knowledge? Going with the current analogy, knowledge in this sense is the ability to reach beyond a single isolated light into the infinite field of consciousness. You might be able to refine and train your hearing and your sight to allow you to study more of reality, but there is still a limit to what you can hear and smell, which means that the reach of the five senses is limited. The reach of the sixth sense, the secondary sense, which is thought, allows us to use language and mathematics create abstractions that can predict what will happen in a galaxy a million light years away from here. In a way, it allows us to create additional senses to explore consciousness and the Universe with. That said, and this is why its a secondary sense, none of this is a matter of direct experience, and that brings with it occasional problems.
Thought and knowledge impose abstractions onto reality, and with the right thought and the right knowledge, they allow us to map this reality fairly well. That said, no matter how good the map is, it’s still a map and not the actual thing. Observations and intuitions through the other five senses allow us to directly experience this reality. There is no map. It’s just a bare, naked experience that connects to the brain. Now, of course, it’s well-known that these other five senses can lead us astray (immediate anger, for example, isn’t always a reflection of the true reality in front of you), but if adequately trained (as contemplative traditions like Buddhism aim to do), then they are a far stronger reflection of a particular lived environment than thought.
It’s no coincidence that advanced meditators, who have refined their senses to a higher degree than people less acquainted with the path, are said to possess a higher degree of wisdom, and that’s because their experience of reality is truer, less clouded. They have learned to directly interact with their surroundings in a way that harmonizes their being with that of the being around them. In this way, we can say that thinking, the secondary sense, is what allows us to build knowledge (which is both collective — creating science — and individual — learning science), and in this way, knowledge errs towards rationalism. But the other five senses allow us to create wisdom, which is only ever individual, and it errs towards empiricism. Reducing one to the other ignores the fact that they are interactive in a way that perhaps we don’t have the vocabulary to fully map.
In this sense, if we take it back to the professor and the fisherman, we can say that the professor has knowledge about the sea, whereas the fisherman is wise in regards to how act in harmony with the sea. This distinction is important because one references a secondary sense (thought) and its ability explain things far beyond the reaches of the other senses (although only in terms of hypotheticals because it hasn’t experienced them) and the other references the five senses that can be refined to understand things well enough to give us information about how to actually act in the world in front of us.
If the professor suddenly went out into the sea with only his knowledge and without any experience, he may have a slightly easier time interacting with the sea than, say, someone who is completely blank, but there is no way that he would have the intuition that adequately tells him how to survive a storm or how to respond to the currents in the right way. Conversely, the fisherman may be able to navigate all of the harshness that this world throws at him, but he can’t tell you why in a way that makes universal sense.
In the field of psychology, the concept of IQ, which is supposed to roughly measure general intelligence (mostly hereditary) has a robust history of research behind it. In fact, it’s one of the most concretely tested measures in the field and the correlations it shows are comparatively sturdy. Yet, there is a lot controversy about whether or not it really plays as big of a role in the real world as is espoused by some people. Naturally, people have an incentive to both downplay its role (“It’s not fair that something so out of our control should dictate so much of what we get out of life”) and to upstage its role (It’s really hard to accurately measure these things, and some people have an undue confidence in establishing correlations as if they suggest something they actually don’t). The question, then, is: How much does IQ matter as it relates to things like success in the real world?
In the framework I have laid out, IQ would roughly capture abstract thinking ability, or the capacity to create and accumulate knowledge. Now, does knowledge help in navigating the real world? Or better yet, is the professor more equipped to deal with the harshness of the sea than the average person? And the answer is clearly yes. That said, a fisherman doesn’t need a high IQ to dominate in his area of expertise if he has spent time accumulating wisdom in that particular domain and correcting for errors over time.
Wisdom can be both contextual (being a great fisherman or being a great soccer player or being a great copywriter) or it can be general (understanding and dealing with life in a healthy way as, say, a monk would be better equipped to do), and both of these kinds of wisdom can be helped with knowledge but knowledge isn’t a requirement for them to manifest if the empirical capacity of the senses in the person embodying them has been developed to a high enough level of competence, and an IQ test has nothing useful to say about that. All it does is tell you that you have the inborn capacity to accumulate and create knowledge, which is clearly important, but not important enough, because the real world goes one step beyond theory, and that is, it requires action — the ability to interact with and adapt to a changing reality, which is an entirely different ball-game.
When a fisherman is out in the sea, he moves with the waves, and he dances with the life-forms beneath him, without thought, without abstractions. He experiences subtle vibrations of physical matter on his body, and his brain then contextualizes these vibrations based on prior experience, which has been earned by prior mistakes and lessons, and it tells him exactly what to without actually telling him anything at all. There is no way to entirely replicate the effect of this process without actually having lived the life he has lived in relation to the sea — no knowledge, no IQ test, can save him without this background of having walked the actual path.
The professor may do important work in the field of oceanography, and this work may even tell us something new and important about our relationship to nature, enhancing our collective knowledge in such a way as to push us towards a brigher future, but this domain is different from the domain of lived experience, different from the subtleties of reality.
This way of thinking has a lot of benefits, but one of the clearer ones, to me, is that it reaffirms the truth of the old cliche that everybody can teach you something. As a naturally curious person — at times, an arrogant person — someone who at a young age had managed to learn a lot, and then learned to use that knowledge to disarm people with word-games, I had always been quick to think that I know more than I do — that if I can intellectually understand the logic of something, I get it; that I don’t always need to hear someone out, nor do I need to respect the wisdom of their experience and what that has taught them.
Now, time is a generous teacher, and while I’m still not completely beyond this kind of thinking, I am getting better at it — I am more eager to pause, to listen, to give people space when they appear to be grasping something they feel is of substance even if they don’t have the language to fully communicate that substance in a way that naturally resonates with me.
It’s easy to take one side of the argument over the other, which we usually do based on our own unique biases and our own unique predispositions, but the truth is that, in actuality, things are messy, and people are complex, and the world we interact with is even more complex. Many different things can be true at the same time, depending on the interactions that are in play, depending on the predominant context. The point is never to declare victory for rationalism or empiricism, or for IQ or no-IQ, but it’s to honestly assess and see what works — why, how, and when.

Tuesday, 15 January 2019

Are we underestimating the harms of legalizing marijuana?

Aaron E. Carroll, professor of pediatrics at Indiana University School of Medicine
Those who hold this view have been in the news recently, saying that research shows we are moving too far too fast without understanding the damage.
America is in the midst of a sea change in policies on pot, and we should all be a bit nervous about unintended consequences.
Vigilance is required. But it should be reasoned and thoughtful. To tackle recent claims, we should use the best methods and evidence as a starting point.
Does Marijuana Increase Crime?
Crime has gone up in Colorado and Washington since those states legalized marijuana. It’s reasonable to wonder about the connection, but it’s also reasonable to be skeptical about causation.
The best method to investigate this may be synthetic controls. Researchers can use a weighted combination of similar groups (states that are like Colorado and Washington in a number of ways) to create a model of how those states might have been expected to perform with respect to crime without any changes in marijuana laws. Benjamin Hansen, a professor of economics at the University of Oregon, used this methodology to create a comparison group that most closely resembled the homicide trends and levels from 2000-12.
“I picked those years because they were after the tremendous crime drop in the early ’90s and most predictive of crime today,” he said. “I ended in 2012 because that’s when Colorado and Washington voted to legalize marijuana.”
This model showed that we might have predicted more of an increase in Colorado or Washington just based on trends seen in comparable states, even without legalization. When he compared the two states with the synthetic control, Colorado and Washington actually had lower rates after legalization than you’d expect given trends.
This is not evidence that legalization lowers crime rates. But it does suggest that we shouldn’t conclude that it increases them. A number of other studies agree.
What About Car Crashes?
A potential misperception involves automobile crashes. Drunken drivers are measurably impaired when their blood alcohol level is above a certain level. We can prove this in randomized controlled trials.
The tests we use for measuring the presence of THC, though, do not measure the level of impairment. They measure whether someone has used marijuana recently. If we legalize the drug, and more people use it, more people will register its recent use even when they are not impaired. So it should be expected that more people involved in car crashes will test positive even if no one is driving while high.
Using a synthetic control approach, Mr. Hansen and colleagues showed that marijuana-related fatality rates did not increase more after legalization than what you would expect from trends and other states.
The Concerns About Schizophrenia
Dr. Ziva Cooper is one of the authors of the National Academies of Sciences, Engineering, and Medicine’s comprehensive report on cannabis.
She says some have misinterpreted the report to state that the report’s committee concluded that cannabis causes schizophrenia. It did not.
“This was stated as an association, not causation,” she said. “We do not yet have the supporting evidence to state the direction of this association.”
Dr. Cooper, research director of the U.C.L.A. Cannabis Research Initiative, went further: “We as a committee also concluded that a history of cannabis use is associated with better cognitive outcomes in people diagnosed with psychotic disorders. The blatant omission of this conclusion exemplifies the one-sided nature of some articles. Nonetheless, the strong association between cannabis use and schizophrenia means that people with predisposing risk factors for schizophrenia should most certainly abstain from using cannabis.”
Marijuana Has Downsides
No one should be under the impression that marijuana is harmless. The potential downsides are well known, and I’ve covered them. Nor should anyone be irrationally exuberant about its upsides. It’s not a wonder drug, and the proven benefits are also minimal (as I discussed here).

We should be honest about what we do and don’t know. We need more research. It’s true that much of the literature around marijuana focuses on the negative, but that’s “largely due to funding priorities over the last several decades,” Dr. Cooper said.
In the report she worked on, only 40 of the 450 pages were about the therapeutic effects of cannabis and cannabinoids, she said, while the other sections were related largely to the negative health outcomes.
She added, “With increased awareness of the clinical potential of cannabinoids, research priorities have shifted to include studying this area” in the last few years.
It’s perfectly natural to be concerned that as cannabis products become legal in more states, they will affect more people.
Many of the experts who have done the work highlighted here are still nervous about how we might proceed. No one thinks that children or adolescents should use marijuana. There’s little regulation right now, and there’s potential for the drug to be mixed with other substances to increase its addictive properties. Advertising will probably make claims that will be out of line with reality.
We should be clear about what we’re talking about here, though. Adults will make decisions on how to use it just as they do with similar products like alcohol and tobacco. Both are more dangerous than marijuana, and it’s not even close.
Anecdotes can make compelling cases, but they don’t necessarily lead to thoughtful outcomes.

Tuesday, 25 December 2018

Rituals can heal

Ari Honarvar
I don’t know if I could have survived seven years of my childhood without the soul-saving rituals of my Persian culture. I grew up amid the Iran-Iraq War, which killed a million people. Besides the horrors of the war, freedom of thought and expression were severely restricted in Iran after the Islamic revolution. Women bore the brunt of this as, in a matter of months, we were forced to ditch our previous lifestyle and observe a strict Islamic attire, which covered our bodies and hair. We lost the right to jog, ride a bicycle, or sing in public. Life seemed unbearable at times, but we learned to bring meaning into uncertainty and chaos by maintaining grounding practices and developing new ones.
It helped that in Persian culture we had ceremonies to turn to. We clung to 3,500-year-old Zoroastrian ceremonies that correspond to the seasons. Several of these rituals take place during the spring because the equinox marks the Persian New Year. Besides a thorough spring cleaning, we jump over a bonfire to cleanse our inner landscape and give our maladies to fire and gain vitality from it. On the longest night of the year, winter solstice, we stay up all night eating fruits and nuts, reciting poetry, playing music, and dancing. This is to symbolize survival and celebration during dark times.
Cristine Legare, a researcher and psychology professor at the University of Texas at Austin, says, “Rituals signify transition points in the individual life span and provide psychologically meaningful ways to participate in the beliefs and practices of the community.” They have been instrumental in building community, promoting cooperation, and marking transition points in a community member’s life. And as strange as rituals might be from a logical perspective, they have evolved as distinct features of human culture.Rituals, which are a series of actions performed in a specific way, have been part of human existence for thousands of years. They are not habits. According to research psychologist Nick Hobson, a habit’s inherent goal is different from a ritual’s. With habit, the actions and behaviors are causally tied to the desired outcome; for example, brushing our teeth to prevent cavities and gum disease and exercising to keep healthy. Rituals, on the other hand, are “goal demoted,” which means that their actions have no instrumental connection to the outcome. For example, we sing “Happy Birthday” to the same melody even though it isn’t tied to a specific external result.
While it’s not clear exactly how they help, rituals reduce anxiety, improve performance and confidence, and even work on people who don’t believe in them, research shows. In a University of Toronto study, participants who performed a ritual before completing a task exhibited less anxiety and sensitivity to personal failure than when they completed the task without first performing the ritual.
Additionally, rituals benefit our physical well-being and immune system. According to Andrew Newberg, the associate director of research at the Marcus Institute of Integrative Health, rituals lower cortisol, which in turn lowers heart rate and blood pressure and increases immune system function.
We live amid a loneliness epidemic where the lack of belonging and community has been linked to high suicide rates and an increased sense of despair. The United States has one of the worst work-life balance scores in the world, while more Americans have become disillusioned with organized religion, as a broad and rapidly rising demographic consider themselves spiritual but not religious. Perhaps with fewer opportunities for people to be in community, many shared cultural rituals are falling away and with them a grounding source for connection and mental health.
“We are an intensely social and ritualistic species.” 
In Iran during the war, we found uses for rituals when we were faced with food rations. We gathered family and friends, reciting the ancient story of the poor abused girl who had run away from home and had a vision of being visited by three celestial bibis (matrons). The bibis instructed her to make a sweet halva and donate it to the poor. The girl said she didn’t have any money, and the bibis told her to borrow or work for the ingredients. This worked well with food rations as each guest brought a few ingredients to make the halva. Like the girl in the story, each participant made a wish and took a bite of the halva. I walked away feeling calmer and more supported.
Stories, such as those told during the Jewish ceremony of Passover Seder, have become ritualized because they are recited in the same way each time. Rhythm and music play a similar role in ritual. Whether we’re chanting in Sanskrit or singing the national anthem, “our brains tend to resonate with those around us, so if everyone is doing the same dance, hymn, or prayer, all of those brains are working in the same way,” Newberg explains. “This can engender a powerful feeling of connectedness. It also reduces stress and depression through a combination of effects on the autonomic nervous system, which is ultimately connected to the emotional areas of the brain—the limbic system.” According to one study, chanting the Sanskrit syllable “om” deactivates the limbic system, softening the edge of fear, anxiety, and depression.
Psychologist Hobson confirms that rituals aren’t just a benefit to our mental health—they’re actually essential. “We are an intensely social and ritualistic species,” he says. “Take this piece out of our modern human narrative and you lose a piece of our history and our humanity.”
I moved to the U.S. when I was 14. After living here for two decades, I became a mother and was confronted with the phrase, “It takes a village to raise a child.” But where was that mythical village and the rituals that made it sane? For example, a pregnant woman in Iran had a rotating menu of dishes made for her by friends and family. A new mother was surrounded by people who took turns assisting with daily tasks. But in the U.S., she was expected to fend for herself and her baby immediately after childbirth. I observed that besides standard holiday traditions, community-building practices were lacking.
So after 20 years of living in the U.S., I decided to create my own community rituals.
I started with my family. At dinners we banned books and devices, lit candles, and discussed set topics of conversation. We held weekly family meetings with opening and closing ceremonies and used a talking stick to enforce respectful communication. At birthday dinners, we took turns saying, “I love you because …”
Candlelit dinners were no longer saved for a special occasion. Using a talking stick helped me listen more attentively and choose my words more carefully. Huddling together at the end of each family meeting provided me with a sense of accomplishment. Each ritual, no matter how small, anchored me in something bigger and provided a sense of belonging.
Then we began to build rituals within the larger community. First, we hosted a multigenerational Sunday potluck with friends and family. Each week, five to 10 of us gathered, shared food, and recounted what made us grateful. During each meal, I noticed I was lighter, more engaged with others, and laughed more.
Later, we built more community rituals into the week. I posted on Nextdoor, asking our neighbors to join us on Monday evening walks to the neighborhood park and back.
In this age of isolation, we need nourishing and uplifting means of creating community by bringing together members of different generations as our ancestors did. From my experience in Iran, rituals can be particularly valuable during hard times. In the U.S., we don’t have to worry about bombs and food rations, but we still have challenges to our security that affect our mental and physical health. Rituals can help us, though, by offering our communities opportunities for healing and support.

Thursday, 6 December 2018

Les algorithmes détériorent la démocratie


 Comment avez-vous pris conscience que les algorithmes pouvaient représenter un danger ?


Cathy O’Neil, mathématicienne et ex-analyste à Wall Street : Je travaillais auparavant dans la finance, en lien direct avec la modélisation des risques. Et ce que j’ai réalisé, c’est que ces algorithmes n’avaient jamais été conçus pour être vraiment corrects : car si les risques étaient sous-estimés, alors nous gagnions plus d’argent. Ça a été l’une des principales causes de la crise financière de 2008.

Vous vous souvenez des titres adossés à des créances hypothécaires, notés triple A ? Cette note était censée dire que, mathématiquement, ils avaient peu de risques de s’effondrer. Alors que c’était faux. Mais les gens y croyaient, parce qu’ils font confiance aux maths.

Est-ce avec cette crise que vous avez compris l’impact que pouvaient avoir les algorithmes sur la vie des gens ?

Je n’ai pas forcément mesuré l’impact à ce moment-là, mais je savais que les conséquences étaient silencieuses et invisibles. Il y a eu un long laps de temps entre les emprunts et la perte des maisons, le lien n’était pas si simple. Et puis les emprunteurs avaient honte : ils ont été pointés du doigt pour avoir acheté des maisons trop chères pour eux. Le mécanisme social de la honte a dépassé le débat sur les algorithmes.

Quel est l’exemple le plus frappant, selon vous, d’« arme de destruction mathématique » ?

Aux Etats-Unis, des juges utilisent un algorithme, Compas, qui évalue la probabilité pour un prévenu de se faire arrêter à nouveau dans les deux ans à venir. Or, on sait que chez nous, les gens se font arrêter pour tout un tas de raisons : s’ils se droguent, s’ils sont pauvres et qu’ils urinent sur le trottoir parce qu’ils n’ont pas accès à des toilettes, ou s’ils sont noirs et qu’ils fument de la marijuana –, les Noirs se font arrêter bien plus que les Blancs pour ça, même s’ils fument autant. Il y a beaucoup de raisons de se faire arrêter qui n’ont rien à voir avec des actes violents, mais avec le fait d’être pauvre ou issu des minorités.

Et les gens que cet algorithme estime à haut risque sont emprisonnés plus longtemps, ce qui, ironiquement, augmente le risque qu’à leur sortie, ils aient moins de relations sociales, donc moins de chance de trouver du travail… Et donc plus de risques de retourner en prison.

Vous n’êtes pas la première à dénoncer les biais des algorithmes censés prédire la récidive. Pourquoi sont-ils toujours utilisés aux Etats-Unis ?

Il y a malheureusement une sorte de division politique entre les personnes qui font confiance à ces algorithmes et les autres. Ce qui est ridicule, puisque c’est censé être de la science basée sur des faits. Mais c’est justement ce que je veux démontrer : ce n’est pas une science, on n’a aucun moyen de tester les hypothèses, et même quand on pourrait le faire, on ne le fait pas !

Qui plus est, cet algorithme, tout comme celui censé prédire où envoyer les policiers pour éviter les crimes, repose sur l’hypothèse selon laquelle nous avons des données sur le crime [sur lesquelles se basent ces algorithmes pour faire leurs prédictions]. Or, nous n’avons pas de données sur le crime, mais sur les arrestations.

Dans votre ouvrage, vous dites que les pauvres sont les premiers à souffrir des algorithmes. Est-ce vrai dans d’autres domaines que la justice ?

Les pauvres sont les grands perdants de l’ère du big data. Quand je travaillais comme data scientist, mon travail était de distinguer les consommateurs à haute valeur des consommateurs à faible valeur. Par exemple, je suis moi-même une consommatrice à haute valeur, surtout pour les sites de tricot, car j’adore ça. Je suis donc vulnérable aux offres sur le cachemire.

Mais les personnes pauvres, elles, sont ciblées par des industries bien plus prédatrices. Comme des universités en ligne qui ne sont là que pour le profit, qui visent surtout des personnes si pauvres qu’elles bénéficient d’aides de l’Etat pour payer leurs études, et qui ne connaissent pas vraiment le système. Ce sont des personnes vulnérables, à qui on dit que tous leurs problèmes vont être réglés par cette éducation en ligne.

Et ce sont des algorithmes qui déterminent votre valeur…

C’est ce que fait le big data, oui. Il s’agit d’utiliser des informations qui ne paraissent pas pertinentes au premier abord, comme des « j’aime » sur les réseaux sociaux, des retweets, des recherches Google, ou bien le genre de site médical que vous visitez… Toutes ces informations sont utilisées pour créer un profil et voir si vous êtes quelqu’un de privilégié dans la vie. S’ils décident que vous l’êtes, alors ces algorithmes feront de vous quelqu’un d’encore plus privilégié. Et inversement. Ils exacerbent les inégalités.

Vous dites même, dans votre livre, que les algorithmes représentent une menace pour la démocratie. Dans quel sens ?

Je pense surtout aux réseaux sociaux qui nous servent de l’information. Le problème est que leurs algorithmes ne s’attachent pas à nous donner des informations justes, mais des informations qu’on aurait envie de voir, en se basant sur les clics d’autres personnes qui nous ressemblent. Ce sont les fameuses bulles de filtre.

Le problème, c’est quand on les combine avec la publicité ciblée. Dans les dernières heures de la campagne présidentielle américaine de 2016, l’équipe de Donald Trump a envoyé des publicités aux Afro-Américains sur Facebook pour les pousser à l’abstention. On essaie de manipuler les gens de façon très personnalisée. Je ne suis pas manipulée de la même façon que vous l’êtes. On pourrait imaginer, par exemple, qu’une publicité me dise que je suis grosse, un jour d’élection, pour me décourager d’aller voter. Et ça pourrait marcher !
Ces algorithmes nous empêchent d’accéder à de bonnes informations, car ils nous manipulent émotionnellement. Ils détériorent la démocratie.

Pensez-vous à l’inverse que les algorithmes puissent être utilisés aussi à bon escient ?

Certains algorithmes, selon leur utilisation, peuvent vraiment aider les gens ou détruire le système. Prenez la santé : vos données peuvent être utilisées par un médecin pour vous dire si vous présentez certains risques, quel traitement vous devriez prendre, etc. Mais le même programme peut être utilisé par des assurances pour se débarrasser des gens qui risqueraient de tomber malades et de leur coûter cher.

Le bon algorithme est celui qui travaille pour l’égalité. Pour cela, il faut se demander à chaque fois à qui cet algorithme nuit.

Vous évoquez dans votre livre une « guerre silencieuse », dont « les victimes n’ont pas les armes nécessaires pour livrer bataille ». Que peut-on faire à l’échelle individuelle ?

C’est une question très importante et, malheureusement, je n’ai pas de bonne réponse. Imaginez qu’un algorithme d’embauche rejette votre candidature parce que vous êtes une femme. Comment allez-vous le savoir ? Comment allez-vous trouver d’autres personnes traitées injustement ? Et comment allez-vous vous rassembler, vous organiser et vous battre ? C’est impossible.

Et du côté des entreprises, des gouvernements ?

Il faudrait déjà appliquer les lois existantes ! Nous avons des lois qui rendent illégale la discrimination à l’emploi des femmes, il y en a aussi sur la finance et la justice, que l’on ne fait pas respecter. En revanche, sur la question de l’information, des bulles de filtre, je pense qu’il faut de nouvelles lois, même si c’est dur de dire à quoi elles devraient ressembler.

Par exemple, Facebook devrait s’ouvrir aux chercheurs. Aujourd’hui, ils en autorisent de temps en temps, mais ils choisissent les projets, ils ont le contrôle sur les expériences et ils décident de la publication ou non des résultats. Ils ne devraient pas avoir ce pouvoir. On a assez de raisons de suspecter Facebook d’affecter la démocratie, il faut donc qu’on sache exactement ce à quoi on a affaire.

On pourrait aussi interdire de cibler les gens avec de la publicité politique. Vous pourriez diffuser des publicités politiques, mais pas décider qui les voit. Tout le monde devrait voir la même chose.

Pensez-vous que la situation va empirer ?

C’est possible. Regardez la Chine ! Des efforts comme le RGPD [le nouveau règlement européen sur les données personnelles] sont de bons signaux, mais cela ne concerne pas directement les algorithmes, seulement les données. La loi européenne est en retard sur ce qui se passe aujourd’hui.