UNESCO advocated ROAM principles for Steering AI for Knowledge Societies

UNESCO’s recent Mobile Learning Week was the context for showing how the ROAM principles can reveal what Artificial Intelligence means for the future of communication and information.

“The summary of research that we are launching here shows why digital development should be aligned with human Rights, Openness, Accessibility and Multistakeholder governance – the ROAM principles,” said Guy Berger, Director for Freedom of Expression and Media Development.

“These four principles underpin UNESCO’s concept of Internet Universality, and they can likewise help to ensure that no one gets left behind by the development of Artificial Intelligence,” he said.

Titled Steering AI for Knowledge Societies: A ROAM Perspective,  the summary of the research done by the Communication and Information Sector was launched in a dedicated panel session on March 5, 2019 at UNESCO headquarters in Paris.  

Within the mandate of promoting freedom of expression and access to information, the research links the development of AI to the wider Internet ecology, and covers each of the ROAM principles as well as cross-cutting issues.

Concerning the Rights dimension, Xianhong Hu (UNESCO) underlined that freedom of expression, access to information and media pluralism risk being undermined by increasing personalisation of information and content moderation by AI. She pointed out that Internet users are not aware of how services platforms continuously collect data and threaten the right to privacy, both online and offline. The right to equality and the right of political participation can also be impacted by smart algorithms and micro-targeting techniques, she said.

“Some machine-learning algorithms have a level of complexity that represent an obstacle to the needed transparency of AI,” said Bhanu Neupane (UNESCO), talking about the Openness dimension. He also stressed that open data repositories play an important role in reducing entry barriers to AI.

Concerning the Accessibility dimension, Prateek Sibal (UNESCO) used the example of a social mission start-up in Africa to illustrate the different factors needed to succeed with an AI-based solution, such as access to research, to education and human resources, to data and to connectivity and hardware.

“Effective multistakeholder processes are: inclusive, transparent, responsive and accountable,” said Guy Berger, referring to the Multistakeholder dimension. They can enrich the ensemble of values, norms, policies, regulations, codes and ethics that govern the development and use of AI, he said.

Crosscutting issues included AI and gender as well as AI and Africa. Concerning the former, Lucia Flores Echaiz stressed that “AI applications, such as voice assistants, can reinforce gender norms and biases since they are developed and deployed in a context that reflects gender inequalities.”

The launch of the research summary was followed by a discussion panel composed of academics, professionals, NGOs and other organizational representatives.

“UNESCO’s ROAM-X approach has the advantage of being flexible. It responds to the need to use values to bring about the potential of AI that is not at the expense of equality, human rights and inclusion,” highlighted Anriette Esterhuysen, consultant with the Association for Progressive Communications (APC).

Commenting on the research, John-Shawe Taylor, UNESCO Chair in AI at University College London, stressed that “the ROAM approach as proposed by the research, is appropriate for AI governance and provides a good level of thinking to address AI in a creative and dynamic way.”

Vice President of iFLYTEK, Shipeng Li, addressed the positive usage of AI in breaking down language barriers. He highlighted that “data issues have been there since the Internet exists. Therefore, AI is not the problem; rather, ‘old problems’ are now affected by AI. The ‘problem’ lies within which actor(s) control the technology.”

A further intervention came from Elodie Vialle from Reporters Without Borders in relation to Rights. “AI can help journalism in many ways, such as software correcting mistakes or databased investigative journalism. However, AI has been used to amplify hate speech and micro-targeted disinformation and could also have negative impacts concerning the protection of journalistic sources,” she said.

“Trust can only be achieved through Open, Accessibility and Multistakeholder principles,” said Nigel Hickson from ICANN. “The technological sector must discuss and address mutual trust and reliance between all stakeholders involved. Everyone must be able to trust that their human rights are respected, and that each stakeholder upholds their mandate and values,” he added.

Echoing this perspective, Alexandre Barbosa from CETIC.br highlighted the importance of the multistakeholder dimension and linked it to the openness and transparency aspects, stating, “A multistakeholder approach allows the AI industry to be transparent in its actions.”

UNESCO’s ongoing research further analyses the changes that advanced ICTs bring to society, using the prism of the ROAM-X framework and the normative principles that anchor inclusive knowledge societies and sustainable development. The final publication will elaborate key options for actions for different stakeholders as well as overarching options for shaping the future of AI in communications and information.

More on the subject in the preliminary brochure as well as on UNESCO’s webpage dedicated to Artificial Intelligence.