Buongionro, don Luca Peyron <dluca.universitari@gmail.com> writes:

A group of international scientists coming from various institutions including Princess Margaret Cancer Centre, University of Toronto, Stanford University, Johns Hopkins, Harvard School of Public Health, and Massachusetts Institute of Technology are calling for more transparency within artificial intelligence (AI) research.

https://www.unite.ai/international-scientists-call-for-more-transparency-in-...

Grazie mille per la segnalazione! Anche il sito Harvard School of Public Health commenta lo studio pubblicato da Nature: https://www.hsph.harvard.edu/news/press-releases/a-call-for-greater-transpar... Lo studio «The importance of transparency and reproducibility in artificial intelligence research» è disponibile anche su Arxiv: https://arxiv.org/abs/2003.00898 --8<---------------cut here---------------start------------->8--- Scientific progress depends upon the ability of independent researchers to (1) scrutinize the results of a research study, (2) reproduce the study’s main results using its materials, and (3) build upon them in future studies​ 2​ . Publication of insufficiently documented research violates the core principles underlying scientific discovery​ 3,4​ . The authors state ​ “ T he code used for training the models has a large number of dependencies on internal tooling, infrastructure and hardware, and its release is therefore not feasible” ​ . ​ Computational reproducibility is indispensable for robust AI applications​ 5,6 more complex methods demand greater transparency​ 7​ . In the absence of code, reproducibility falls back on replicating methods from textual description. Although, the authors claim that ​ “all experiments and implementation details are described in sufficient detail in the Supplementary Methods section to support replication with non-proprietary libraries” , ​ key details about their analysis are lacking. Even with sufficient description, reproducing complex computational pipelines based purely on text is a subjective and challenging task​ 8,9​ . --8<---------------cut here---------------end--------------->8--- (i numeri del testo che ho incollato fanno riferimento alle note a piè di pagina dell'originale) Questo problema ovviamente non è limitato alla ricerca che fa utilizzo delle tecniche c.d. AI ma è esteso a tutta la ricerca che fa uso di dati elaborati tramite software... credo quindi di poter dire con un discreto livello di accuratezza che è un serio problema di tutta la ricerca scientifica. La bella notizia è che ci sono TUTTI gli strumenti tecnici per affrontare il problema, se lo si vuole affrontare :-D Cordiali saluti, Giovanni. -- Giovanni Biscuolo