Social changes - Technological changes: a crossroads of views

Numerous innovations are revolutionizing the world, calling into question economic balances and lifestyles. Technological advances and the social transformations that result from them or are at the origin of them give rise to both enthusiasm and controversy, both on their long-term scope and the risks they entail. Changes in work and employment are a matter of continuing concern and attention, and digital transformations sometimes accelerate earlier ones. What about human adaptability to progress? How can contemporary changes be approached "at the same time" from both a technological and a social point of view, by cross-referencing the views of researchers? The role of digital platforms, the place of the State and the transformation of public action, the contributions of digital technology to health and citizen participation, the consequences on forms of work, the lessons of history to put the apparent novelty into perspective, all these points constitute the major questions of our future. The seminar "Social and technological changes" organised by France Stratégies will shed light on this future through the analyses of researchers, entrepreneurs and actors in the field.

 

 

Big data and artificial intelligence

 

Massive data collection, known as big data, multiplied computing capacities and artificial intelligence techniques are the main drivers of the technological transformation at work. Even if artificial intelligence is still far from its ultimate goal of reproducing human intelligence, its progress and potential, discussed during the seminar by Jean-Pierre Ganascia, professor at the Pierre and Marie Curie University, and Yann LeCun, director of Facebook Artificial Intelligence Research (FAIR) and professor at New York University, have largely transformed society in recent years [2]: the web, biometrics, image and speech recognition, and route finding are daily examples.

In the 2000s, a new wave was observed with the arrival of machine learning, which allows computers to be trained to perform a task based on examples, and no longer just to program them to perform a task.

In the 2010's, artificial intelligence has benefited from the progress of new deep learning techniques, thanks to which automatic learning can be applied to more complex tasks such as image recognition. An emblematic application of these techniques, pioneered by Yann LeCun, is driving assistance, with vision and perception systems that enable a car to detect obstacles or change lanes automatically.

Empathetic social robots, humanoid or not, are another application - presented during the seminar by Laurence Devillers [3], professor at Paris-Sorbonne University - of these learning machines developed thanks to deep learning algorithms. Gradually introduced in our homes or hospitals, they are programmed to capture, interpret and respond to emotional information.

The next big research challenge is unsupervised learning, to teach machines common sense and improve, for example, virtual assistants, of which Google Now and Siri are the first models.

These new technologies, the quantities of digital data available and their potential to create value have called into question economic balances, favouring the rise of intermediation platforms such as Google, Amazon, Uber or Airbnb. These platforms, which ensure a personalised relationship with the consumer, without any physical link, derive their wealth from the size of user networks and today benefit from a quasi-monopoly situation. Stéphane Grumbach, Director of Research at Inria, and Laurent Gille, Professor of Economics at Télécom ParisTech [4] presented the challenges of these new balances during the seminar.

Do the transformations at work constitute a new industrial revolution? Pierre-Cyrille Hautcoeur [5], President of the EHESS, highlights a number of points in common with the industrial revolution of the 19th century as analysed by the historian Jan De Vries (Stanford) under the concept of "industrial revolution": intensification of work, increase in the number of hours worked, stagnation of real wages, transfer to the market or mechanisation of activities previously carried out at home and internationalisation.

It is therefore in the transformation of lifestyles and consumer expectations as much as in a supposedly autonomous evolution of technologies that we must look for the drivers of innovation. Sylvain Allano [6], Scientific Director and Future Technologies at PSA Peugeot Citroën, presented the developments in the automotive sector, emblematic of the second industrial revolution. For him, the players, jostled by new entrants such as Tesla Motors, have already entered a paradigm shift linked to the arrival of composite materials, advances in sensors and digital processing capabilities that make cars more autonomous. This paradigm shift can also be seen in supply management, which is less and less aimed at consumers and more and more at users of mobility services looking for vehicles adapted to shared uses. Finally, it is reflected in work organisation methods, in favour of open innovation and highly qualified workers.

The changes in the labour market brought about by new technologies occupy a large part of the public debate, which was widely echoed by the speakers at this seminar. The development of artificial intelligence and robots raises the question of the transformation of professions and the complementarity between man and machines. But if alarmist forecasts evoke a threat of substitution of nearly 50 % of jobs within twenty years [7], they are not corroborated by studies on tasks or taking into account the economic incentive to replace men by machines. According to a note published by France Stratégie [8], this rate would be closer to 15 %. However, the adaptation of skills to this new environment and thus the role of initial and continuing training are likely to be key issues for future thinking.

The growth of platforms is also accelerating several trends in society. While Frédéric Mazzella, founder and president of Blablacar, and Marie-Anne Dujarier, senior lecturer at the Sorbonne-Nouvelle Paris 3 University, are looking at existing platform models [9], Ursula Huws, professor at the University of Hertfordshire, and Antonio Casilli, senior lecturer at Télécom ParisTech, are looking at the emergence of a "cybertariat" [10].

To identify the effects of the platform economy on the labour market, Marie-Anne Dujarier has developed a typology of platforms [11]. 11] Two factors differentiate them: whether they are commercial or not, and whether they are for-profit or not. Thus, for-profit market platforms favour the domination of the owners and designers of the platform, who capture a large part of the wealth produced and rely on a mass of non-salaried micro-entrepreneurs.

Conversely, contributors to for-profit non-market platforms generally do not feel that they are in business, as do drivers registered on Blablacar. These platforms may, however, compete with existing professional activities. According to the sociologist, these two models are constantly questioning the boundary between amateurs and professionals, as well as the very concept of "work" in our institutions. The absence of a defined status and a system of solidarity, as well as the weakness of structured social movements in these new forms of work, calls into question our models of social protection.

For Ursula Huws, who introduced the term "cybertariat" [12] to describe the emergence of new forms of work, the arrival of the platform economy has caused an imbalance in favour of capital. The basic model of wage-labour has been undermined by a new global division of labour, by pro-business political representation and by the economic crisis, whose effects on the precariousness of labour have reinforced the platform economy model. The most precarious forms of employment resulting from these global transformations, invisible to our eyes, are the digital workers, studied by Antonio Casilli, paid by task or by click from developing regions, who feed the algorithms put in place by the major digital platforms, the most emblematic of which have the acronym GAFA (Google, Apple, Facebook, Amazon).

 

 

How do we interact with machines?

 

While the development of digital interfaces raises apprehensions, particularly about the transformations brought about in the labour market, it nevertheless meets strong user expectations. Sophie Pène [13], professor at the University of Paris-Descartes, mentions a 2008 UK Post Office study according to which 53 % of English mobile phone users are "nomophobic", i.e. live in fear of disconnection, without their mobile phone. The ubiquity of these interfaces in our daily lives raises the question of our interactions with machines and their effects in different areas of our lives.

The research carried out agrees on one point: there are as many impacts as there are technologies, uses and even people.

Thus, according to Daphné Bavelier [14], professor of cognitive neuroscience at the University of Geneva, regular practice of combat video games can improve selective visual attention. And if the smartphone has made us lose declarative memory, it has on the contrary made us gain transactive memory (ability to search for information).

 However, it seems that the positive effects disappear with hyper-specialisation in one area. The long time required for evaluation limits the consideration of these effects in public policies. According to Roberto Casati, Director of Studies at EHESS, the "1 million tablets" plan in the 6th and 5th grades was implemented without a prior study of the impact on school performance and, above all, without any thorough reflection on the use of digital technology in teaching practices.

In the field of health, experiments using sensors and connected objects on patients or the introduction of social robots in the management of certain pathologies have given initial results on the effectiveness of these tools. For Sophie Fainzang [15], research director at Inserm, the key to the success of sensors in improving prevention, medical monitoring and patient autonomy is the construction of a collaborative relationship and the redefinition of the role of patients and their doctors. A passive use of these tools lacks the assigned objectives: only 50 % of chronic patients equipped with sensors systematically transmit their data.

According to Anne-Sophie Rigaud [16], head of department at Broca Hospital, the acceptability of social robots with dependent people varies according to the desired use: patients who want simple assistance in daily life will prefer robots without emotion, while those who are looking for cognitive stimulation or a reassuring presence will prefer robots with empathy.

Bruno Sportisse [17], deputy general manager of Thuasne [18], insists on the importance of trust in connected objects and their use.

As far as the business model of the platforms is concerned, confidence is again the key to success. According to a study conducted by Blablacar and NYU Stern [19], we trust a platform user almost as much as we trust a friend.

Such a relationship is rather paradoxical because the underlying algorithmic processes are beyond the understanding of most individuals who do not master either the trajectory of the data produced or used, or the computational logics of these devices. In this context, the key notions of freedom and responsibility are questioned. Researcher at the University of Namur, Antoinette Rouvroy insists on the difference between classical machines and algorithmic machines [20]: programmed to orient our actions by giving us information and recommendations, the latter can even substitute for human decision making.

The advent of social networks as an information channel is another example of the risks of uncontrolled algorithms for accessing information. Dominique Cardon, professor at Sciences Po, and Serge Abiteboul, research director at Inria, warn about this evolution [21], which is less "neutral" than Google's historical ranking algorithm, which is rather transparent and "meritocratic". By analyzing our preferences, these filtering algorithms and the choice of social network by users reinforce the informational bubbles and the virality of certain "fake news".

During the seminar, on various topics, several speakers referred to this lack of control over algorithms, but also more broadly to the issues of data ownership, transparency and cybersecurity. What limits should be given to uses, what functions do we want to assign or delegate to these algorithmic machines, what degree of opacity are we willing to accept? All these questions that revolve around responsibility are still largely unanswered.

 

For public authorities, digital development is seen as a means of improving public services at lower cost. However, there is a risk of expecting too much. Thus, civic techs, presented by Stéphanie Wojcik, lecturer at the University of Paris-Est Créteil, and Clément Mabi, Vice-President of Démocratie ouverte, are often seen as a possible response to the "democratic fatigue" that plagues traditional institutions and modes of participation [22]. 22] Public authorities regularly mobilize them for information and consultation purposes. But the legitimacy of the results is questionable if these tools are essentially invested by politicised publics.

For Clement Mabi, they are not a panacea for the civic divide. What is more, their widespread use to access public services can reinforce the exclusion of groups that combine social and digital precariousness. Pierre Mazet, researcher at the Observatory for the non-use of rights and services, and Jean Deydier, director of Emmaus Connect and WeTechCare, both highlighted the lack of support and progressiveness of the dematerialisation of public services [23].

In addition, democratization initiatives to promote empowerment (theempowerment) of citizens can lead to disappointment. Experiences in France and abroad show the limits of using digital contributions in the decision-making process, the most flagrant example being the rejection by the Icelandic Parliament of a new constitutional law drafted by Internet users.

Weaknesses in digital regulation were mentioned several times during the seminar. The history of technological risks, studied by Jean-Baptiste Fressoz [24], a science historian at the CNRS, teaches us that it was the social controversies caused by accidents due to the introduction of steam engines and gas lighting that gave rise to the main principles of regulation (prior administrative authorization, standardization of technical devices and compensation of victims in the event of damage). Applying these principles to the digital sector to prevent computer bugs and security breaches would already be a major step, as recommended by Gérard Berry, professor at the Collège de France [25]. The lack of social analysis of risks and responsibilities in a digital world increases the likelihood of an inappropriate takeover by a company or a State, of instrumentalization and unwanted or undesirable use of data gathered by machines. The a posteriori response is not enough. Thus, the Hadopi law came to regulate the practice of downloading at a time when it was disappearing in favour of streaming. According to Daniel Le Métayer, director of research at the Inria Grenoble Rhône-Alpes research centre in Lyon, transparency must be considered from the very beginning of the service's design by those who process personal data (privacy by design). For Laurence Devillers, it is also upstream that the autonomy and decision-making capacities of the robots should be defined by ensuring the transparency of the data collected and the algorithms used.

The last point of convergence of most speakers in these debates is the crying need for education and acculturation of the population to digital tools and the underlying mechanisms. In the face of the upheavals brought about by digital technology, in the labour market or in democratic participation, the education system has an essential role to play. For François Taddéi [26], Director of the Centre for Interdisciplinary Research, this system is changing little or not at all, even though the relationship to knowledge and the role of the teacher are undergoing major changes. This observation is shared by Roberto Casati [27]: digital technology seems to him to be used more in a mimetic mode, to reproduce what already exists, than for its own virtues such as learning the code, creating content or sharing data.

 

 

The fourteen meetings organised in two years within the framework of the seminar "Social and technological change" are far from having explored all the themes. They are fleeting and localized spotlights to approach this vast matter in motion that is our society, driven by the social and technological forces that it incorporates, digests and absorbs.

At the end of this exploration, the questions remain. What will be the future of the work? What place for France and Europe in the face of the digital giants? What regulation for this new world? What horizon for public action? Beyond these questions, convictions are being forged, drawn from the many echoes between sessions: the importance of data management in which users have a firm grip; education about tools, which should not be confused with usage; refusal to give in to sirens or prophets, and the desire to take the right measure of the transformations under way. The world is changing; we must adapt the rules by knowing the grammar.

 

2 See Session 7, " The place of man in the face of artificial intelligence".

3 See Session 8, " What place for robots in society? ».

4 See Session 2, " Big data: value creation, challenges and regulation strategies".

5 See Session 1, "The digital "revolution", an industrial revolution? ».

6 Ibidem.

7 Frey C. B. and Osborne M. A. (2017), The Future of Employment: How susceptible are jobs to computerisation, Technological Forecasting & Social Change No. 114, pp. 254-290.

8 Nicolas Le Ru (2016), "The effect of automation on employment - what we know, what we don't know", France Stratégie, La Note d'analyse, No. 49, July.

9 See Session 3, "Collaborative economy, economy of sharing: what are the challenges for tomorrow? ».

10 See Session 12, " Cybertariat".

11 Dujarier M.-A. (2016), "Social uses of digital technology: a typology".In Digital labor, travail du consommateur, quels usages sociaux du numérique, INA Global, La Revue des Industries Créatives et des Médias, January.

12 Huws U. (2001), "The Making of a cybertariat? Virtual work in a real world"Socialist Register 2001: Working Classes, Global Realities, vol. 37.

13 See Session 10, " What are the impacts of the screens on the brains? ».

14 Ibid.

15 See Session 5, " The doctor-patient relationship in the age of connected objects".

16 See Session 8, "What place for robots in society? ».

17 See Session 5, "The Doctor-Patient Relationship in the Age of Connected Objects.".

18 At the time of his intervention.

19 Blablacar, NYU Stern (2016), Entering the trust Age.

20 See Session 6, " Algorithms, freedoms and responsibilities".

21 See Session 11, "Dissemination of information: what are the democratic issues? ».

22 See Session 9, "Is digital technology transforming citizen participation? ».

23 See Session 13, "Poverty and Digital".

24 See Session 4, "Technological Risk in the Digital Age".

25 Ibid.

26 See Session 14, " How are teaching methods evolving? ».