The need for such an ethical code arises because Artificial Intelligence (AI) is an increasing feature of political campaigns, for instance by ‘profiling’ audiences or to target the most important voters with tailored messages. With this paper, Bakir wants to bring the message to policymakers, that they should focus on enhancing transparency and explainability, which would better equip voters to assess campaign messages to which they have been exposed. An ethical code can help to achieve this.

Financing and profiling

While there are benefits from digital political campaigns, such as the potential to better engage hard-to-reach parts of the electorate or making politicians more responsive to electorates, digital political campaigning is currently very opaque, resulting in possible harm. Such campaigns could negatively impact citizen’s ability to make informed choices or fragment national conversations, thereby weakening our ability to hold political campaigners to account. Bakir illustrates two major issues in the current reality of online campaigning, with examples from the Brexit campaign, proving the need for an ethical code:

• There was opacity concerning the breaching of spending limits, coordination of campaigns with other groups and sharing of data among the several “Leave” campaigns.
• There was also opacity about what data-mining services the political campaigns actually used. Use of such tools, and such data sharing, to optimise messages for profiled audiences is especially problematic when the messages are emotive, deceptive and targeted at audiences predisposed to be receptive to such messages.

Bakir fears that these issues are likely to get worse with the increasing use of AI in political campaigning. She discerns at least four harms to democracy. The use of AI in political campaigns can

• negatively impact citizen’s ability to make informed choices;
• fragment national conversations and limit citizen’s ability to hold political campaigners to account;
• increase potential for targeted voter suppression;
• enable exploitation of people’s psychological vulnerabilities.

Ethical code

An ethical code is needed to reduce the harms of AI in political campaigning. What elements need to be included in this code in order for it to be effective?

• Transparency: Make clear if political messages online come from a political party, how much campaigners spend on digital campaigning, and on what.
• Explainability: In campaigns that extensively use AI to profile voters, give all voters an explanation of the profiling.
• Civility: Campaign material should be civil (e.g. not nasty, aggressive, disrespectful, or pitched to provoke anger and outrage) and must not incite others to commit crimes (e.g. making false statements of fact about candidates’ personal character or conduct).
• Informativeness: Campaigns should give voters enough information to freely make informed judgments. The information provided should be true, complete, undistorted and relevant.

The views expressed in this paper are those of the author and do not necessarily reflect the views or policies of the Knowledge Center Data & Society or CiTiP. The paper aims to contribute to the existing debate on AI.