Only AI support for customization

In Italy, to become a university professor it is necessary to have the national scientific qualification (also called ASN). It is not a sufficient condition and currently thousands of researchers have it in their possession without being able to use it. Current legislation stipulates that the process for awarding the ASN is managed by a committee made up of all full professors who meet strict restrictions on their scientific activity. More specifically, given the diverse composition of the Italian academic system, 190 committees are currently active, one for each competition area. In a speech held on June 23 at the Crui (Conference of Rectors of Italian Universities), the minister of the outgoing government, Cristina Messa, announced the abolition of national commissions and the use of “indicators for requirements and automated control (keywords and Artificial Intelligence IA )”. The immediate reactions of the scientific community led to a clear reduction of these proposals by the minister herself.

Is the case closed? Not really. In recent days, the Board of Directors of the Italian Association for Artificial Intelligence (AIxIA), with the support of the Board of Directors of the Group of Computer Engineers (GII) and the Board of the Group of Informatics (GRIN), published a note that “aims to analyze the limits of the specific application and feasibility “of the proposal of the Minister of Masses (https://aixia.it/comunicato-su-asn/). In short, the memo recognizes the limits of AI in performing such complex and human tasks (the evaluation of ASN candidates), placing this activity among those defined as “high risk” by the current proposed regulation under discussion to the European Commission. It was, however, particularly the conclusions of the note that re-opened a heated debate and sometimes very harsh tones. “In summary, and more conceptually, AI techniques could be used, even in the short term, to develop increasingly powerful and reliable indicators: indicators in any case, not value assessments,” he says.

It stands to reason that the computer academic community and, in particular, that dealing with Artificial Intelligence, sees the potential to contribute to the execution of such a complex task and, by its very nature, with subjective and nuanced elements. Science aims to tackle challenges that seem insurmountable and certainly cannot be overcome. It is also plausible that this possibility is framed in a very rigid and regulated normative, but also moral and social framework. In fact, science has an obligation to anticipate the social impact of its discoveries, as it can never be neutral in its applications.

What cannot be accepted is the expectation that the development of decision-making tools to support the work of national scientific training committees will take place in a short period of time, because there would be no room for vigorous and in-depth discussion at various levels ( academic, scientific, social and political) about the impact of this approach. Moreover, the acceptance in such a sensitive context of the partial replacement of human judgment with that of an artificial system (whatever it is, intelligent, automated, digital, cybernetic) would open the way for other replacements, in contexts perhaps less “shady” but equally important.

In 1968 Philip K. Dick published “Are Androids dreaming of electric sheep?’ (But Do Androids Dream of Electric Sheep?, in Italy also published as “The Android Hunter”) inspiring Hampton Fancher and David Webb Peoples to write his play Blade Runner, the 1982 cult film directed by Ridley Scott. In the fantasy world of many of us, Artificial Intelligence is synonymous with robots, androids and replicants. Even if we know that this is not the case, the pressing demand for objective, strong and refutable guarantees cannot be ignored.

Leave a Comment