Today is not possible to find a unique definition for discrimination, as a phenomenon. Notwithstanding, the scholars has formulated some guidelines to identify when we are in front of discrimination. For example, Professor Costanza Nardocci understands it as conflicts produced and created among ethnic and racial groups that get in contact with each other; or as a tool with which the dominant group can maintain the status quo and be in charge of the scarce resources; or as a result of social and collective prejudice.
Because of this ambiguous conception, we can track the origins of discrimination in social conflicts produced by political, ethnic, or economic reasons. Usually, when there’s a social conflict between two or more groups, the society is divided in two, the dominant group which oversees the resources, and the subdued minority or minorities.
Under the former stratification, discrimination plays an important and essential role, it serves as a tool for the dominants to maintain the power over the others and over the economic and/or natural resources.
How does discrimination work? Basically, because of the diverse reasons that can create a social conflict; there are a variety of factors of discrimination that can be found. For instance, race, economic status, ethnicity, and gender, have been the classical factors identified and used to stratify the society. But now, with the evolution of new technologies, new factors have been introduced that makes even more difficult for people to identify when they are being discriminate. As I will revise later, right now Artificial Intelligence is in the scope because of this.
Important is to highlight that, when we talk about minorities or subordinate groups, we shall not only considerer the quantity factor, i.e, there is not a direct relationship between domination and number of individuals who constitute that dominant group. Here is when I would introduce the gender factor.
Historically, women have been treated as a secondary group relegate to the private sphere, in charge of the domestic job. The female empowerment is a young movement that started a couple of centuries ago, but that has been strongly confronted by the dominant social group under this factor -gender-. Even though women have represented half of the population since the beginning of humanity, is common knowledge that the opportunities and the role played by female individuals is completely different that the one played by their male counterparts. This is why number is not synonym of domination.
This situation is replicated in any aspect of the society, women have had access to education, political positions, economic resources, independence, decades later than men. Consequently, the influence and participation of women in any of these areas still is highly disproportionate.
Today we are experiencing a wave of tech evolution, especially with the development of Artificial Intelligence -AI henceforth. The problem is that with the introduction of new technologies, new situations and factors of discrimination have been found too. For example, with the face recognition systems, have been reported several cases of error in identifying African women, even confusing them with animals -as it happened with Michelle Obama. Or new recruitment systems that have incurred in discrimination when choosing a possible candidate for the job.
But what is AI? Like what happens with the definition of discrimination, is impossible to find a unanimous definition for Artificial Intelligence. For instance, I will refer to the definition given by the EU High-Level Expert Group on Artificial Intelligence, as a “software (and possibly also hardware) systems designed by humans that, given a complex goal, acts in the physical or digital dimension by perceiving the environment through data acquisition, interpreting the collected structured or unstructured data, reasoning on the knowledge, or processing the information, derived from this data and deciding the best action(s) to take to achieve the given goal”.
Therefore, Artificial Intelligence systems collect data from the external world and interpret this data by reasoning and processing the information. So, under this perspective, how could AI systems discriminate? Is not directly the machine in charge of discriminate per se, but as the definition says, the problem is in the data recollection phase.
AI technologies are fed by data provided by humans. If the information given by individuals is biased, the result provided by the machine will be biased too. Here plays importance the power relationship between the dominant group and the subordinate one.
As I referred before, women have been historically relegated to a secondary place, entering to the public spaces and education sphere later than men. This phenomenon increases in technology-related areas, as UNECO’S Report “I’d blush if I could” informed, the presence of women represents 22% of AI professionals globally, less than 1% of applications for technical jobs in Silicon Valley come from female candidates, women assume only 15% of top-level positions at leading technology firms, and women are 25% less likely to know how to leverage ICT for basic purposes. Hence, is very common to experience discriminative results during the interpretation process because the lack of data regarding female physiognomy or personal aptitudes, or even the knowledge that women have of these technologies.
On the other hand, this is not an intent to affirm that women, right now, are more victims than men regarding AI discrimination. Quite the opposite, is just a description of one of the new forms of discrimination born with new technologies that specifically affects women, without excluding other possible new factors that affect men differently and make them be discriminated the same.
What are the possible solutions to this special problem? One thing can be empowering women to participate, study and work in the STEM disciplines to acquire more influence during the process of technological evolution. Other solution can be found in making noticeable the current situation, so male and female scientists and people of interest can consider the discriminative scenario that new technologies have created and take care about these gaps.
Either way, the Artificial Intelligence evolution is here to stay, so the only measure that we can take right now as society, is evolve along with it.
References
https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
https://unesdoc.unesco.org/ark:/48223/pf0000367416.page=1
chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://unstats.un.org/unsd/gender/downloads/Ch8_Poverty_info.pdf
chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://nces.ed.gov/pubs/96768.pdf
https://ainowinstitute.org/publication/discriminating-systems-gender-race-and-power-in-ai-2
