UNESCO has released a report titled “Artificial Intelligence and Democracy”. The report delves into the current and potential impact of artificial intelligence on democracy and the benefits that both artificial intelligence and digitalisation, in general, could bring to enhancing collective decision-making processes.

The analysis is structured on the democratic expectations and disappointments of digitisation, new digital public space: the democratic conversation, democracy of data: the politics of Big Data, and study democracy as a form of political decision-making: algorithmic governance.

According to the report, the development of digital technologies, data, automation, and artificial intelligence has sparked great expectations and fear. Beyond concerns regarding employment, rights, and freedoms, there is apprehension about their potential impact on political life and democracy. 

In the report, democracy is regarded as the structure of social coexistence and a governing system crafted throughout history to facilitate the free self-determination of individuals. Notions of representation, participation, legitimacy, and solidarity underscore a humanist vision of societal structure that aspires to make individual rights compatible with communal responsibilities. Genuine democracy only exists where citizens are well-informed, have open channels for participation and advocacy, and have a say in the decisions that affect them. Considering all of the above, while holding considerable promise, current AI technologies also harbour significant risks, several of which this report addresses

The digital space

“Democracy hinges on robust conversation and an environment conducive to it. Digitisation is leading to such a radical modification of the public space that it forces us to reevaluate how this dialogue—essential to democracy—unfolds. We must assess how to enhance it, identify its vulnerabilities, and address potential threats”, the study writes.

UNESCO, through the report, suggests that we must now explore why what was once hailed as an open environment without owners turned out to be a restricted space with gatekeepers. While digital giants have successfully positioned themselves as neutral intermediaries, their claim to neutrality is contradicted by their control over access and usage and their influence on content through recommendation algorithms. 

If  they  want  to  act  in  line  with  democratic values,  platforms  must  comply  with  strict  transparency  standards  by  disclosing  the  inner  workings  of  their  algorithms that sort out information, suggest posts, or assess sponsored content.

The role of data

Governance based on available data is not neutral or  indisputable  because  the  data  itself  is  not  neutral  or indisputable.  We  must  analyse  the  new  power  dynamics  that  result  from  the  analysis  of data. Big Data is  as  much  of  a  political  issue  as  data production, distribution,  and consumption,  where access, control,  and  resources  and  capabilities  are  disproportionately  distributed  due  to  uneven  power dynamics.

Furthermore, data is not politically neutral; data collection, analysis, and use hinge primarily on specific decisions. The  more  data-driven  policies are,  the  more  crucial  it  becomes  to  scrutinise  the  explicit  or hidden assumptions in data selection or the biases they manifest.

UNESCO Recommendations

Based on the study, UNESCO put forward a set of recommendation for democratic governance of AI.  Mentioned below are the key recommendations of UNESCO. 

• Member States should ensure a balanced and preventive narrative, allowing citizens to understand the true implications of technological transformations and feel protected by their authorities
• It is important to establish mechanisms for dialogue with national and regional parliaments, mainly through science and technology committees. 
• To ensure maximum transparency, essential for an informed public opinion, States are encouraged to promote codes of  best  practices  for  companies  and  require  identifying  products  generated  by  artificial intelligence as measures to combat misinformation.
• It  is  vital  to  acknowledge  that  much  of  our  democratic  citizenship hinges on data regulation, ownership, and use. Member States should regulate this issue within the scope of their competencies. Given the nature of data, adopting a novel approach to manage it is critical.
• As the number of decisions that are made, in whole or in part, by automated decision-making systems increases, democratic principles need corresponding procedures to regard algorithmic decisions as democratic. Encouraging transparency and explainability in artificial intelligence systems is crucial to making it easier to understand decision-making processes and the criteria underlying their outcomes.  
• The  regulation  and  governance of AI  must  be  complemented  by  institutions  facilitating  public legitimisation,  which  is  necessary  for  this  normative  principle to be enforced.
• Member States should adopt national digitalisation and artificial intelligence strategies with adequate budgets. These strategies should encompass goals related to technological transformation and economic modernisation as well as others with direct democratic implications.
• Member States must ensure citizen engagement, oversight, and independent assessment of AI and data protection systems.

Full report