Minutes from Adria Information Disorder AI Tools 2018 Workshop, held on June 29-30, 2018, in Koper, Slovenia.
Andrej Školkay opened the meeting, welcoming and introducing the others physically present – John Boyd, Bence Ságvári and Francesco Adolfo Danza, (later arrived Peter Matzneller and Michal Hradický) and those online (Juraj Filin, Nikolau Stefanos, Bernie Bencoe, Boris Zizek).
He went on to talk about how Facebook (FB) is developing its own AI, employing some 20,000 people to combat security and content, also in the name of fake news.
Francesco interceded to note that FB had decided to close out a lot of things (in respect of data access) making it much harder to monitor certain processes. He believes news could and should be take out of FB altogether. John agrees that it may be the best solution, albeit infeasible really.
FB claims its AI and software are succesful at eliminating fake news, but it is not in the business interest of FB to ban fake news, as it is not focused on truth, noted Franscesco.
Andrej continues, stating how there are already around 50 orgasnisations in Europe tackling the problem of fake news and about 150 worldwide. So how to get more deeply involved?
According to the research of SCM, the software Fighthoax claims to have the best overall score, with the company claiming its algorithm works at 89% accuracy. Claimed to be the first AI algorithm to check and detect FN.However, comparative as well as independentent individual assessment shows that Fighhoax, which gained a lot of publicity within EU, does not seem to perform so well. This inconsistency, as well as the general curiosity about involvement of AI into fighting FN, as presented in policy recommendations on fact-checking fact checkers, carried out within COMPACT Project in early 2018, was the main reason to organise this small workshop.
