Between 8 and 9 November, two representatives of the Internet & Society Reference Institute participated in the Global Symposium on Artificial Intelligence & Inclusion, organized at the Museum of Tomorrow (Rio de Janeiro) by the Institute of Technology and Society of Rio de Janeiro (ITS Rio) and the Berkman Klein Center for Internet & Society (Harvard). The event’s proposal stems from a larger platform, the Global Network of Internet & Society Research Centers (NoC), of which IRIS is part and which has more than 80 centers of technology in the world.
Theme of the event
The symposium was attended by guests and collaborators from around the world in various fields: law, philanthropy, media, psychology, education, politics and industry. The objective was to address the main opportunities and challenges of artificial intelligence-based technologies by means of the inclusion perspective. Artificial intelligence and its related technologies have begun to shape important parts of the digital economy and affect key areas of our society, which is increasingly connected. Whether it is for the transportation (self-driving cars), industrial production, finance, social justice, health or education sectors, AI has the potential to profoundly impact our lives and shape our individual and collective future in ways we do not even we currently conceive.
The promise of AI-based technologies is huge and the benefits can range from efficiency gains to unprecedented improvements in quality of life. Potential challenges and risks are equally astonishing, for example, when one considers uncertainty about the future of work or the emergence of new power structures beyond the control of existing governance and accountability frameworks. More specifically, unequal access and the impact of AI and related technologies on often marginalized populations, including poor urban and rural communities, women, youth, LGBTQ, ethnic and racial groups, people with disabilities – and particularly those at the intersection of these marginalized groups – contribute to the disruptive risk of widening digital inequalities around the world.
In addition, it was emphasized that those interested in the area of AI should take the methodological care to seek to distinguish what are the new issues and problems that arise with the new technology, problems and issues already existing in other areas of knowledge. This point is important so that the debate and resolution of problems is carried out in an efficient way, avoiding a generalized hype about AI.
Artificial intelligence and exclusion: recurrent marginalities
The production of inequalities by artificial intelligence was a recurring theme of the symposium. For example, the bias of results in surveys in image banks was highlighted, through the work of the collective Desabafo Social. In addition, we discussed the ways in which infrastructure, data selection and the training of application developers can reproduce (sometimes even more exacerbated) the inequalities that processes of marginalization that we already observe in “offline” society.
One of the keynote speakers at the symposium, Nishant Shah, explained that often the decisions made by developers and even the legal teams that help in the management of these applications are based on the concept of “average man” or “reasonableness” of what is customary and socially expected.
However, the very concept of “average man” already carries diverse biases and presumptions. For example, it would carry the idea of a white, western man from the Global North. The technology involved in processes that presume the existence of an “average man” attitude, or situational reasonableness, in fact, is not neutral. One one hand, assuming that this technology will not be neutral, why not think about what would be appropriate, adequate and ideal biases for its development? How about trying to include also through artificial intelligence?
On the other hand, artificial intelligence also requires contextualization. In Japan, for example, the figure of robots and the automation of thought has been part of the collective imagination for decades, but in a very positive way. According to researcher Arisa Ema (Tokyo University), artificial intelligence is usually treated positively in animes, films and reports. Instead of an apocalyptic Terminator-of-the-Future scenario, AI applications range from dementia treatment, depression, and even learning facilitation methodologies.
An important issue raised at the Symposium is the difference in the ability to develop AI technologies when comparing developed countries with the Global South. Today few countries have been leading technological development in this area, especially in the USA and China. This concentration raises concerns as it has the potential to aggravate inequalities between countries. Given that the possibilities for a greater universalization of IA production by several countries is a barrier to the concentration of data by certain actors. This is important because, in order to train an AI algorithm, in many cases, a large amount of data is needed, which has focused on some economic agents like Amazon, Apple, Microsoft, Google, Facebook, and Baidu, for example.
Another important criticism made in the Symposium was that certain AI algorithms can be trained only with data from a specific language and culture, which can generate problems in the use of this algorithm when applied to another language and culture.
Unfortunately Brazil is far behind when compared to the leading countries in the development of this technology. Vladimir Putin has been trying to put Russia in this technological race, in national public class opening the school year said that Artificial Intelligence is the future of humanity, and that those who lead their development will lead the world. The tone of speech, though exalted, should serve as an alert for our country. In this sense it is interesting the text of the anthropologist Hermano Vianna that plays with the concept of anthropophagy of Brazilian Modernism stating that we must develop an anthropophagic AI to the Brazilian:
“Alright, a lot of what is currently said about AI is hype. But hype generally creates such “windows of opportunity” that can be taken advantage of by those outside the main debate, using the general blah-blah-blah mess to give pitaco in the conversation we were not called to. In my opinion Brazil missed the opportunity of the internet. We did not invent an original application for the use of the Great Network, like Skype created by Estonians, Swedes and Danes … We did not even create local versions of global tools that could then impact the rest of the world, as the Chinese did with the Alibaba or the Baidu. The Brazilians were basically satisfied with the role of voracious consumers of news, champions or vice-champions in use of social networks […] Our behavior was increasingly acquiring the standard of voluntary servitude before big corporations like Facebook. We exchange the freedom of open source (or the notion that information wants to be free) by closed condominiums run by smarter companies in the use and sale of our private data. I suspect that when it comes to national development, most Brazilians (including Brazilian economists) are thinking only of buying Abercrombie shirts via Amazon.
My consolation: There may be something creatively predatory in this consumerist trend. Is there a Brazilian way of thinking? Could we here construct a really artificial other intelligence, alien to dominant patterns, or outside the Putin project of world domination? Brazil is not over?”
Copyright and artistic production
Another interesting aspect of the symposium concerns discussions about copyright and artistic production through artificial intelligence. Harshit Agrawal, an Indian artist, produces installations that somehow involve creative automation through AI. In one of his plays, volunteers were called in to hold a pen on a paper, which was then commanded by a magnet. An AI made the pen move and produce sentences and content on the paper. Part of the discussion of the work involved further discussion with these volunteers, more specifically about the feeling of paternity they felt about the “created” works by their hand. The objective is precisely to discuss the pertinence (or not) of the authorship of artificial intelligence. According to the artist, he himself does not see enough elements to mark the authorship between the AI programmer and the works he creates.
Want to know more about artificial intelligence? Check out some of the posts we have already written on the topic here on the blog, as well as the Global Symposium on Artificial Intelligence & Inclusion’s website. The results of the conferences will be published shortly.