ARTIFICIAL INTELLIGENCE VS. BIOLOGICAL INTELLIGENCE
Listening to Laurent Alexandre once in a lifetime is an almost mystical experience. At CREA, we have been following him for several years with fascination. With no taboos, no tongue in cheek and sometimes with a touch of Darwinism, this extra-lucid on the upheavals linked to Artificial Intelligence, and more generally to new technologies, takes us out of the ambient European torpor in which we are vegetating and confronts us with the reality of the international situation, American and Chinese first and foremost. In the manner of an Incal, far from the soothing or distressing speeches, he casts an acid and frank look at the future that awaits us.
Summary of a conference given by Laurent Alexandre to students last June at the Inseec Lyon Campus.
Interview by Illyria Pfyffer
PART 1 HUMAN, TOO HUMAN.
What are the risks and opportunities for biological intelligence from the massive advent of Artificial Intelligence?
En termes de business, il faut arrêter de parler d’intelligence artificielle seule, ce qui compte c’est
In terms of business, we must stop talking about artificial intelligence alone. What counts is the mix between artificial intelligence (AI) and biological intelligence (BI). We have never seen a microprocessor start reading Kant without reason. Artificial intelligence does not exist without biological intelligence. And conversely, a manager without artificial intelligence will soon be unemployed. We need to think about the sum of the two intellects, which are intimately linked and cannot be separated. Artificial intelligence is first and foremost biological intelligence that is industrialised, scalable and packaged. Instead of talking about intelligence, it is better to talk about the management of intelligence. The economic world will find itself permanently faced with the need to interface AI and BI. This is the main job of the 21st century: to organise the fusion between AI and BI. As intelligence is the source of all power, what happens in this field has consequences for both the organisation of companies and power. The issue of the AI/IB couple is therefore major. Will we prefer a society of intellectual solidarity, or will we move towards an intellectual apartheid – between ‘gods’ who master AI and useless people with miserable incomes who are waiting to die? It will depend on what the new generation decides.
Another difficulty facing the economy is that it has industrialised AI before it democratised and promoted IB. The first time I wrote an article on the decline of the IQ level in Europe, it sparked an outcry, now it is a certainty and daily newspaper Le Monde has just published an article on the subject. The IQ of Europeans has dropped by 4 points over the last 20 years, while at the same time Asians have gained 10 points. We are therefore witnessing a European decline in the stock of biological brains. This creates a gap between the decline in our IQ, for whatever reason, the explosion of AI and the explosion of Asian IQ.
“What matters is the mix between artificial intelligence (AI) and biological intelligence (BI). We’ve never seen a microprocessor start reading Kant for no reason. ”
“This is the main job of the 21st century: organising the fusion between AI and BI.”
What are the benefits and limitations of AI?
AI that travels at the speed of light is ubiquitous and immediately updated. But when you wake up in the morning, your brain has not been updated! Whereas AI has instantly loaded, for example, the changes in labour laws in South Korea. This huge difference makes Yann Lecun, Chief AI scientist at Facebook, say that we will soon realise that human intelligence is very empty. The current stage consists of promoting AI *deep learning. The next step will be to develop AIs with a capacity to learn with small DATA sets, but this is not yet certain; it may turn out that AI will continue to require a mountain of data. Thus nothing allows us to ensure that in 2040 AI will be able to recognise a giraffe when it has already seen four. Today we are faced with two types of AI, symbolic AI (i.e. we code perpetually) and connectionist AI which took off in 2011. A connectionist AI does not invent anything, it relies on the knowledge base with which it is educated. But it is sensitive to trolls, as Microsoft’s Tay* has bitterly experienced…
For the moment, it is unlikely that we will be able to make AIs that learn with little data, as babies do, before 2030. There are only 10 AI producers on Earth, and it is a centralised production that includes the GAFAMs and the BATXs, an American-Chinese oligopoly where Europeans are non-existent.
Some go even further and wish to transcend the limits, like Masayoshi Son, the CEO of the Japanese telecoms group Softbank, the richest man in Japan who bought Google’s Boston Dynamics robots and made the cover of The Economist. He has raised two 100 billion funds this year to develop the Singularity, i.e. the moment when AI will surpass the IB. His goal is to make robots with an IQ of 10,000 within 30 years (the average human has an IQ of 100…).
Deep learning1 is a set of machine learning methods that attempt to model data at a high level of abstraction using articulated architectures of different non-linear transformations. These techniques have enabled significant and rapid progress in the fields of sound or visual signal analysis and in particular facial recognition, voice recognition, computer vision and automated language processing. (Source: Wikipedia)
It took barely 16 hours for Tay to go off the rails. However, everything had started well. On Wednesday, Microsoft’s research unit launched its artificial intelligence project: Tay, a chatbot that thinks it is a 19-year-old woman. It runs on an intelligent program in which Microsoft has included basic knowledge such as bits of sentences. But Tay’s main learning curve is through its interactions with people on Twitter. “The more you chat with Tay, the smarter she gets, so her experience can be personalised by you,” the Washington-based firm wrote in an explanatory note. This option did not leave users unmoved, as they rushed to their keyboards to teach her lots of things. The experiment had to be deleted in less than a day after the writing of racist and neo-Nazi comments on Twitter. (Source: Paris-Match)
PART 2 UNEQUAL INTELLIGENCE AS A SOURCE OF INEQUALITY
What about Europe in the AI race?
In the battle for AI, the forces at work are the United States and China. Europe has slowly become a “digital colony” of GAFAM. Didn’t Total hand over its oil operations to Google last month? And didn’t Carrefour make a deal with Tencent in Asia and with Google in Europe just recently? Thalès announced that it was building a secure cloud for the French army, but in reality it was with the help of Microsoft… Our dependence and fragility in the face of the digital giants remains extremely strong. Between the Washington/GAFAM axis and Beijing/BATX we are altar boys, in the process of technological vassalization. This enormous concentration of wealth, NBIC technologies, fintech, brains, and the most profitable companies in a few metropolises on a global scale is worrying. We are taking a great leap backwards: we are creating an increase in inequality that we have not had for 1,000 years. We are also risking the return of political violence, stemming from these inequalities created by this world where you are going to be either a lord or one of people in a situation of increasing precariousness, especially after the arrival of AI-enabled robots in 20 years’ time.
“Between the Washington/GAFAM axis and Beijing/BATX we are altar boys, in the process of technological vassalization.”
How then can we resist the oligopolistic digital giants?
AI is encapsulating the world’s knowledge and best brains. This industry is currently causing global competition for the best. Do you know how much the very good AI specialists in GAFAM are paid? Down to the last 10,000 euros?
Answer: 2 million per month!
Recently, one of the heads of research in France said: “We are hiring researchers with 11 years of higher education in AI for less than 3,000 euros a month.” If it remains on this trend, Europe no longer has a chance, as tomorrow’s companies will be subject to a race of speed, to new risks, and the GAFAMs could turn us into prey quite quickly. How can we cope with this global brain war that promises to be terrible? Small entrepreneurs who cannot afford computer scientists risk being wiped out. The current landscape resembles the giants that are stalking small prey, either to buy them out, which is a lesser evil, or to butcher them.
Today, it is estimated that there will be a shortage of 85 million highly qualified employees worldwide. In 15-20 years you will be fighting in your companies and start-ups to have qualified workers. Faced with this global shortage, the ultra-rich Chinese companies will do as the GAFAMs have done, and do their business in Europe, especially in French research institutes, since we pay our researchers a pittance.
Is hyper-regulation a solution?
The company of tomorrow will be more political and highly regulated, contrary to what we expect. The economy will become increasingly affected by geopolitical and philosophical implications. We are even going to have to manage hyper regulation because hyper technology begets hyper regulation. Drones in particular will be hyper-regulated. Don’t forget that it has been shown that a small drone can contain 3 grams of Semtex, the same Czechoslovakian explosive that blows up planes. So states are not going to let people do AI in their cellars, synthetic biology in their kitchens, etc. Regulation will be fundamental and constantly increasing.
We are also faced with another difficulty: some people today are convinced that they can give AI good ethics. We just forget that our ethics are constantly changing. Firstly, very complex systems like deep learning are very difficult to audit, and secondly, from a moral and ethical point of view, we act as if moral and ethical standards were universal. There is a certain gap between the morality of Daech and European morality, just as there is a gap between today’s morality and that of yesterday. These very strong differences raise the question of what ethics to give to AI? The balance between measure (hybris) and excess (nemesis) is likely to be complicated. Many people envisage a dystopian future where revolution threatens. So politically, if we want to avoid a 1794-style revolution, we will have to reduce intellectual inequalities and imagine a real neuro-revolution.
How would you describe this AI revolution?
We are facing GPTs (General Purpose Technologies), in other words, excessively violent technologies that are changing not only the technology, but the entire economy. We have had several technology waves with strong implications: steam, electricity, electronics and now AI. AI is particularly potent because, in addition, it allows neuro-technology in the genetic field to change our very identity. It is more disruptive to change the brains of our babies, or to make a genetically modified one à la carte, than to build a faster or slower TGV. The current wave of technology is particularly morally disruptive.
AI will disrupt the lives of individuals and create a complex world, the lives of companies and nations will be marked by unpredictability and the management of inequalities. People without degrees are sinking in our countries, we can’t abandon them. At the individual level, even the head of Google recently said that he was not at all sure that human beings would be delighted to see AI galloping so fast. At the corporate level, with the increase in unpredictability, the advent of the platform economy (Uber, AirBnb, etc.), the rise of exponential technologies, etc., the core of the business will be to manage the interfaces with AI and all the technologies that come with it. But this world is quite oligopolistic, the GAFAMs have no competition, it’s why they make money. We are now in a digital fog with an exponential increase in risk, the beta. Who knew that Kodak or Nokia would collapse? Who would have imagined on the day Netflix was born, that it would be worth 75 times more than TF1, the leading TV channel in Europe. TF1 is only worth 2 billion, a lot of peanuts to Netflix. For those who like utopian scenarios, I invite you to watch the press conference that the boss of Netflix gave. In it he says that selling mp4 files is not his future business model. What he wants is to provide viewers with a pill that will make them experience the film, like in Total Recall. Driving the frontier between AI and IB, organising the synergy, that’s where the best and most rewarding professional opportunities lie!
What fascinates you most about AI?
One of the most incredible phenomena, which no one would have ever imagined, is that in 10 years we all have a second job in addition to our official activity. In addition to our “normal” paid job, we work for free 5 hours a day… to educate the AI of the digital giants! We have been working every day since childhood, for hours on end, to educate the artificial intelligences of the digital giants, by providing them with photos, elements about our personality, our consumption, etc. This is a radical change. In the global economy, there are not billions of people who are willing to work day and night to enrich a business they know little about and from which they derive no benefit. Yet we, billions of useful idiots, work for free to enrich GAFAM. This is unfair competition.
On the scale of humanity as a whole, I am fascinated by the prevailing desire to create a homo deus, as described by Yuval Harari, with demiurgic powers. This concept is being developed by Silicon Valley, by China, towards the exploration of the cosmos, to merge our brains with AI, to develop the baby on demand. The Times explained that “the death of death” could happen as early as 2045. Elon Musk and Neuralink explain that it is urgent to put chips in our brains so that robots do not turn us into servants…
In addition to our “normal” paid professional activity, we work for free 5 hours a day… to educate the AI of the digital giants!
Your futurology essay on the destiny of AI?
You have to be a bit modest, futurologists say anything. Uncertainty is the key word for the years to come. Between the development of AI and the increasingly rapid decision-making process, economic uncertainty will become the rule. How will society react to the challenges of these demiurgic technologies that will have a massive impact on business, now that is the question! We are always smarter afterwards… I won’t have the cruelty of going back to the 1997 New York Times article* which explained that the computer would not be able to beat man at the game of Go for 100 to 200 years, whereas it was already beaten to the punch in 2016. We don’t know how society will react to such disruptive technologies that have such a strong impact on business. Think of the big chains Walmart, Carrefour and others, who when Amazon first started out, thought it would swiftly sink…Today Amazon is a giant and its entry into the space domain, totally unpredictable, is disturbing many. Stéphane Israël, CEO of Arianespace, recently told me: “There is no guarantee that Ariane will still be alive next decade. AI is not arriving on an empty beach, it is arriving on a thousand years of history, which again is a source of unpredictability.
*It may be a hundred years before a computer beats humans at Go – maybe even longer,” said Dr. Piet Hut, an astrophysicist at the Institute for Advanced Study in Princeton, N.J., and a fan of the game. “If a reasonably intelligent person learned to play Go, in a few months he could beat all existing computer programs. You don’t have to be a Kasparov.” (Source: NYT 1997)
PART 3 THE CURRENT ECONOMY AND THE CHALLENGE OF IA
How will businesses be impacted by AI?
Beware of catastrophic predictions. When you read a study that says that 47% of jobs will disappear because of AI, it is only counting the jobs that are likely to disappear, not the ones that are created. If that study had been published in 1895, it would only have counted the number of farriers who would have to be retrained, not the automated jobs that would spring up in the 20th century, from microprocessor maker to heart surgeon to webmaster. In 1915, the New York Times published an article on the lie of the automobile industry, saying in effect: there have never been so many horses in the United States, we have 23 million. So you can see that the car salesmen are lying to you, the car does not replace the horse. “? And yet the demand for horses collapsed in the months that followed…
What is certain is that the production function between capital and labour will be disrupted by AI. I won’t be as pessimistic as Geoffrey Hinton, a leading deep learning specialist in Toronto, who recently wrote two papers suggesting that we should stop training dermatologists and radiologists because AI will perform better than them. When you have a bachelor’s degree, which is the case for a radiologist, and the AI analyses scans better than you do, you don’t stand there like an idiot and become impoverished, you try to become complementary to this AI. All scenarios are still possible, between corporatist blockages, replacement of doctors by nurses, extension of the scope of work, etc. (doctors could also extend the duration of the consultation, including undressing/redressing, from 6 min 30 to 9 min, that would be humane). There is nothing, right now, that says that we will need less, as much or more dermatologists in 2080 than today. And this observation is reproducible in all production functions. What is known in economic theory is that when a new, more efficient factor of production comes along, companies competing with that factor see their value plummet. So the highly skilled labour needed to manage, to encapsulate AI, will see its value rise. It will quite radically change the way we work now. And there is great uncertainty about robotisation. Will the Boston Dynamics robots that everyone admires on YouTube continue to cost $1.5 million apiece for a long time to come, or will Son bring the price down, as he hopes, to $30,000?
We are living in a rather curious period where highly qualified work is at equilibrium, while we have seen a drop in the purchasing power of middle managers over the last 30 years, and an explosion of unskilled work, possibly carried out by people with four or five years of higher education. This is even more evident in the USA, where part of the unskilled work is done by people with degrees. This is not a healthy system. In a second phase, we can fear a drop in demand for medium-skilled work, and if the price of robots drops with democratisation around 2030 and 2040, unskilled work, after having reached a breaking point, will collapse. To make a fortune in 2040, it will therefore be necessary to know how to manage interfacing with AI. As long as AI needs human intelligence, the core of value management will remain the management of this interface.
In the labour market, many scenarios are possible: either we have a ban on AI, or a gentle adaptation, or a merger, but it is not certain that people will agree to implant chips in their brains to have an IQ of 200… I leave it to you to imagine your preferred scenario.
What advice would you give to the younger generation?
I am a surgeon, a doctor, and I have been writing about intelligence for a long time. Many teenagers, students, or even parents, all influenced by AI conferences, TEDx etc., call me and say “My goal is to change the world”. Young people should not set unrealistic goals, otherwise they will become unhappy or bitter. We have to be careful not to encourage the younger generation to become depressed.
We are living in an extraordinary century, the golden age of entrepreneurs, innovators, managers, scientists, etc. So don’t listen to all the talk about the future. Don’t listen to all the old fogies who tell you that the world of AI is going to be dreadful, you mustn’t live your life like Romain Gary lived his life: catastrophized by old age, afraid of everything. The world that is coming is an extraordinary world. You are the most blessed generation in history, you are going to undertake fantastic achievements. Enjoy life, set up lots of businesses, and don’t listen too much to the mortifying utopians who tell you that tomorrow everything will go wrong. You have, I repeat, a fantastic chance, AI will make you the people who will have the most exciting life that humans have ever lived.
A must read: LA GUERRE DES INTELLIGENCES by Laurent Alexandre. Editions JC Lattès