Artificial Intelligence in Knowledge Societies: A ROAM Approach - Open Data and AI

The session “Open Data and AI” organized within the framework of “Principles for AI: Towards a Humanistic Approach?” on 5 March 2019 requested UNESCO to continue leveraging its convening power to increase awareness around artificial intelligence and big data,  support development of inclusive policy on Open Data and support upstream and downstream capacity enhancement.

The workshop noted Data as an essential element for the development of artificial intelligence. The availability of large amounts of user data through services on mobile phones and internet of things among other sources, has led to a variety of AI applications and services. However, there remain many challenges. These challenges encompass issues of access, privacy, discrimination and openness. Several of these challenges are within UNESCO’s mandate of building inclusive knowledge societies for peace and sustainable development.

Ms Dorothy Gordon, Chair of the Information for All Programme at UNESCO pointed out that “despite the fact that we have a huge interest from many donors, we do not seem to have done very much systematically to prepare African countries to have useful data ... [and] in a searchable format that can be combined with other sources to ... yield something [beneficial]”. She stressed the need to bridge gaps in terms of the availability of legacy data, setting policy standards, and enhancing capabilities of people to work with local data sets.

Ms Constance Bommelaer, Senior Director of Global Internet Policy and International Organizations at The Internet Society underlined ‘data commons’ as an interesting solution to explore but one that needs a nuanced discussion around ownership and privacy. She highlighted the need to challenge existing notion of competition and a need for “reconsideration of market values and monopolies”. Stressing the importance of access, she shared the findings of a joint study carried out by ISOC and UNESCO that showed how a combination of local language content and better access policies results in immediate economic benefits at the local level.

As a government representative, Ms Veronika Bošković Pohar, Deputy Permanent Delegate of the Republic of Slovenia to UNESCO discussed ‘regulatory sandboxes’ as a means to provide controlled environment for AI. She hoped that Slovenia’s proposed Category 2 Centre on Artificial Intelligence would be able to make several informed decisions, provide insights into technology and societal interface and create mechanisms for continuous monitoring and reporting to reduce risks posed by AI to vulnerable groups.

Speaking as a panelist representing a knowledge organization, Prof. Maria Fasli, UNESCO Chair in Analytics and Big Data at University of Essex noted the lack of understanding on AI and Big Data and expressed concerns for the difficulty faced by the academic community in accessing data collected by large technology firms for research purpose. She further highlighted the need for high quality representative data to ensure that algorithms are not biased.

Given their experience in tracking innovation trends across the world. Mr Marcus Goddard, Vice President of Intelligence at Netexplo Observatory underlined that “access to data is a necessary but not sufficient condition for innovation. Pointing out the general trends in openness, he mentioned that openness is not Silicon Valley’s top priority and convenience seems to be the norm when it comes to launch of new products and services. He highlighted that even as data is being used in smart cities to improve access and sustainability, it is also increasing the threat of surveillance.

Mr Philippe Petitpont, Co-founder of Newsbridge, a Paris based AI and Media startup, presented the scale of the data problem that the media faces today. He remarked that media companies are gathering 30 million hours of video content every year, a number that does not include social media videos. In this situation, extracting useful insights from these videos is a cumbersome task albeit one that can be performed by AI. They try to leverage AI to help journalists process large amounts of data at lower costs.

The session brought the viewpoints of multiple stakeholders to the discussion table and some of the key concerns included were:

• Urgent need to increase awareness around artificial intelligence and big data;
• Developing strategies to strengthen access to data for training machine learning algorithms;
• Supporting both upstream and downstream capacity enhancement to leverage data for benefit;
• Involving private sector actors in the discussion around access to data and data monopolies; and
• Creating systems for addressing discrimination and biases originating through data and algorithms.

The panel members congratulated UNESCO for facilitating important discussions around issues of rights, openness, access and multistakeholder participation in the governance of data and hoped to engage with the organization for further development of issues around Open Data and AI.

For more information please contact: Bhanu Neupane