Cultural coherence of data: the Pacific speaks up - 15/09/2039
For more than thirty years, the advent of artificial intelligence, and in particular deep learning, has come about through the exploitation of an immense number of data obtained, more or less legally, via the major platforms of the Internet.
The recurring problem with these data, like social science research, has been the fact that most of them were collected from relatively small samples of the world's population.
In fact, this has given rise to many discriminating situations, to say the least, where information technology tools have clearly forced many people around the world, not from the cultural background of these databases, to fit into the frame.
The fact is that with the explosion of artificial intelligence services and tools from deep learning, biased or limiting situations for many communities have become real sources of harm for a huge number of people: inaccessibility to university scholarships, difficulties as for immigration or security checks, total lack of visibility and therefore persistence of social prejudices on these communities.
After a long period of passivity, cultural NGOs and the ministries of education of the Pacific recently met in a seminar to study the situation, try to measure the real impacts of this lack of representativeness of AI databases and make proposals to international bodies.
The objective of this seminar is to give the voice of the Pacific to the many research institutes, companies and governments of states at the forefront in the development of AI, to try to convince them to adopt a more inclusive methodology and integrate the Pacific institutes into their research and development programs.
While research in AI is still relatively underdeveloped in the Pacific, this is not the case for research in linguistics and “language processing” which are very often used in the management and exploitation of databases by artificial intelligence.
As it stands, given the significant linguistic diversity of the Pacific, many academics currently consider that exploiting the diversity present in the Pacific would allow AI research to make significant progress in learning language and voice recognition.
What are the hopes of this seminar? Very weak to be honest, the Pacific is little weight against the technological giants. But, clearly, there is no risk in trying and behind the scenes, many of the participants, although realistic (not to say cynical), nevertheless hoped that this would create a global movement which could, in the long term, weigh more heavily on the “big states”.