The Politics of Machine Learning
University of Helsinki, Finland
Machine learning platforms have emerged as a new promissory technology that some argue will revolutionize work practices across a broad range of professions, including medical care. In 2017, a group of Oncologist at Copenhagen’s Rigshospitalet piloted the IBM Watson for Oncology platform only to realize that they were in full concordance with only 30% of its recommendations. Other studies, supported by IBM, claim concordance rates as high as 96%. In this presentation, I will look at machine-learning platforms, which support medical decision-making in cancer oncology and contemplate the significance of concordance rates as manifestations of evidence. If there is a high level of concordance between the platform and the oncologists, then there is no need for the platform. If there is a low level of concordance with the platform then the platforms decision become suspect. I suggest that there is a way out of this dilemma by better understanding how concordance and discordance are generated and what it says about the decision-making processes themselves. Furthermore, I suggest that concordance levels relate more to knowledge production regimes, rather than absolute truth claims regarding treatment options
Artificial Intelligence as a Disruptive Innovation from an Organizational Perspective
University of Trieste (Italy)
The increasingly pervasive introduction of A.I. systems is one of the distinctive features of the so-called Capitalism 4.0. Analyses of this phenomenon are rapidly flourishing. One branch of studies is dedicated to the possible developments of these technologies in relation to the multiple fields and ways of application, considering the new skills needed as well as the implications for the nature and quality of work. A second large area of study addresses the large-scale effects of new technologies, considering their employment, work and social impact: for example, the possible effects of substitution, creation, displacement and transformation of work, or the effects on the distribution of employment and wealth, on welfare state systems and labour regulation.
This contribution addresses an intermediate analytical area between the two mentioned above, reflecting on the possible effects of A.I. on organizational structure and processes, using some analytical tools provided by the sociology of the organization. The hypothesis that we want to examine is that the introduction of management and operational devices based on the A.I. can constitute a disruptive innovation also from the organizational point of view. This is possible because A.I. interacts widely with the dynamics of learning, communication, decision and action that operate at various levels in organizations. It can therefore be important to reflect, taking a sociotechnical perspective, on how A.I. interacts with different organizational aspects, such as the system of roles and relationships, coordination mechanisms, decision-making processes, authority structures, organizational cultures, etc.
Artificial Intelligence: Where Were We?
University of Turin (Italy), Italy
In the ‘80s the interest in Artificial Intelligence and its promises spread within the social sciences and the sociology in particular. The most ambitious ones explored the possibility of simulating the social behavior with a degree of strong equivalence, that means that it allows to understand the simulated object; from this point of view AI is a cognitive tool.
Today a significant portion of the discourse on AI seems to have abandoned that aspiration. In return, AI penetrated in the reality of our daily lives in different forms and applications; in its name, revolutions of different kind are described or foreseen, some of which also in the practice of sociological research.
"Artificial Intelligence" is nowadays a recurrent expression in different fields, to refer to the most varied applications – that’s why it is often included in buzzwords lists. We try to reconstruct its recent fortune also in sociology, to find the uses of it that refer specifically to simulation – that is to say that aim to reproduce some aspects of the social life through computational models; and to understand if and to what extent the simulation aims to be an explanation.
Artificial Intelligence: A Macro-Sociological Approach
University Niccolò Cusano, Italy
Artificial Intelligence includes two main fields of research: Big Data and Automation (i.e. Industry 4.0 and Internet of things).
Now, sociological scholarship about Big Data is quite large and well established in international academia.
On the contrary, so far sociology has not focused on Automation, even if there is general agreement on the fact that, in the next years, Automation-related innovations will heavily affect many aspects of social life.
Within this framework, a (macro) sociological approach to Automation studies is necessary, in order to find how to cope with the social consequences of the coming productive innovations.
Namely, massively spread Automation-related innovations will be likely to reduce the amount of socially necessary work - both manual and intellectual- and, at the same time, impose a much stronger mastery of automation technologies to all active citizens.
In other words, Automation could be a historic occasion for social progress, while, in the case of bad management, Automation might cause troubles of unemployment, income distribution, and social discontent.
My paper will propose a theoretical model for dealing with this issue, especially:
- an analysis of the main sources for sociologists to achieve reliable information about the actual state-of-the-art of Automation technologies.
- a general pattern for scenario analyses on the Automation impact; this pattern, will include the technical, economic, political and cultural points of view.
- a general pattern for Automation related policies.