Human rights must govern artificial intelligence (AI) in order to address the phenomenon of the lack of transparency of algorithms and the data that this technology feeds on.
The New York-based research center, Data & Society, recently released a document titled Governing Artificial Intelligence: Upholding Human Rights and Dignity, in which it presents some examples of how artificial intelligence can violate people’s human rights, in addition to offering an overview of what the members of society who come into contact with this technology can do about it: companies, civil society, governments, the United Nations, intergovernmental institutions and academia.
“For AI to benefit the common good, at least its design and deployment must prevent damage to fundamental human values. International human rights provide a solid and comprehensive formulation of these values.”, it is mentioned in the document.
Subscribe to our Newsletter
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.
Examples of the consequences of not putting human rights at the center of the development of artificial intelligence projects are not few. Recently, Amazon had to get rid of an artificial intelligence assistant who made hiring its human resources easier because it began to manifest clear discriminatory biases against women.
The organization ProPublica, and especially the researcher Julia Angwin, have been in charge of revealing how the algorithms are “black boxes” that can introduce racist biases in judicial decisions and in an article published on October 21, 2018, the MIT Technology Review exposed how various groups of human rights activists in the United States have spoken out against the creation of “a massive database containing the names and personal data of at least 17,500 people suspected of being involved in criminal gangs.”
The Data & Society document presents examples of how, through artificial intelligence projects developed and implemented by companies like Facebook, governments like China and even universities like Stanford, they can involve systematic human rights violations such as the right to non-discrimination, to privacy, political participation and freedom of expression. “Stanford University researchers trained a deep neural network to predict the sexual orientation of their study subjects, without obtaining their consent, using a set of images collected from online dating websites.
Beyond several methodological shortcomings, the research showed how a lack of respect for the right to privacy increases the risks of algorithmic surveillance, with which the data that is collected and analyzed threatens to reveal personal information about users. This may put individuals and groups at risk, particularly those living under regimes that would use such information to repress and discriminate, ”the report explains.
In what way are companies and organizations worldwide addressing this problem?
One of the companies that are implementing this kind of best practices and recommendations around artificial intelligence systems is IBM, through its Trust and Compliance program, which is available to customers of the company’s cloud service and which has the ability to deliver an indicator report that will tell you how its artificial intelligence platform is behaving from two points of view: on the side of the creation and development of algorithms and from the data set that has been used to train these algorithms.
Although Trust and Compliance are only aimed at customers of the IBM cloud service and universities, this company program stands out against the measures taken by other technology companies such as Google, which also faced claims from its own workers who agreed to discontinue the work in an artificial intelligence project called Project Maven.
“Just a few months later, many Google employees feel that those principles have been pushed aside with an offer for a $ 10 billion Defense Department contract. A recent study was done at North Carolina State University also found that asking software engineers to read a code of ethics does nothing to change their behavior, ”says the MIT Technology Review report.
If AI researchers, developers, and designers work to protect and respect fundamental human rights, they could open the way for broad societal benefit. To ignore human rights would be to close that path.
References
HAO, K., Establishing an AI code of ethics will be harder than people think, MIT Technology Review, https://www.technologyreview.com/s/612318/establishing-an-ai-code-of-ethics-will-be-harder-than-people-think/
LATONERO, M., Governing artificial intelligence: upholding human rights & dignity https://datasociety.net/wp-content/uploads/2018/10/DataSociety_Governing_Artificial_Intelligence_Upholding_Human_Rights.pdf
