White Mirror - The Accountability Machine
In the "White Mirror" series I talk about positive ideas for using digital technologies in a way that can improve our lives. The title is a reaction to the popular "Black Mirror" series of often dystopic visions of our digital future. While highlighting potential dangers is important, shining a light on the positive potential of digital technologies seems to me at least as important.
We've previously talked about the potential positive aspects of using AI in government (see here). And while more and more politicians not only start talking about AI but also continue to use it in their work - just as their constituents - I want to talk about an underexplored use case of AI in politics that could help rebuild trust in democracy in times of a severe media crisis.
The issue
In a democracy, power is directly linked to accountability. Not only are there checks and balances and separation of powers to ensure that people wielding power don't abuse it, the ultimate accountability for politicians is to voters, who placed their trust in them at the time of their election. They will hold politicians accountable to their promises made during the campaign. Media formats and investigative journalists watch the track record of elected officials and regularly provide for an often unwanted reality-check from the politicians' perspective. That's the theory at least.
Cynics might say that this has never worked, yet even idealists would probably agree that the last decades have seen accountability becoming more of an issue. You can sit out scandals and trust an ever faster moving news cycle to overwhelm audiences and wash away your misdeeds in an avalanche of new scandals involving other people.
Digital technology hasn't helped. From a fragmented information ecosystem and social media to AI-assisted deepfakes, the mechanism of accountability is further undermined. And it is also under attack from a different perspective: delegating power to automated decision-making systems strips away the last bits of accountability and hence trust in democracy (remember "Computer says No"?)
The solution
But that is not to say that technology couldn't help. After all, the threat to accountability comes less from technology itself, than humans not properly using it. Why not use AI to help where humans are currently failing? Why not build an accountability machine? Programmes of political parties, campaign promises, press statements on legislative priorities, everything can be tracked and fed into an AI that can act as a point of reference whenever the promises made are brought up. Voters would still be the arbiters of whether promises made = promises kept but it would be harder for shameless politicians to sneak away from being held accountable for their promises.
You don't have time to sift through dozens of contradictory statements by politicians on various platforms? Your editor won't give the resources to analyse campaign promises or ask a politician questions for fear they'll stop talking to your network? Those arguments don't matter in AI-assisted accountability.
An AI can cut through the bullshit, identify patterns, spot differences between what was promised and what was delivered. Don't get me wrong, I'm not advocating for a technocratic dystopia with "delivery scores" for every politician, nor am I saying you shouldn't be able to change your goals once you are in office.
The point of the thought with the accountability machine is more to get us thinking about how we can productively use digital technologies for the better in our political system. It is clear, that the state stands to benefit from introducing AI. But more careful thinking on where and what for AI could be used in the state and political processes is needed. Just as with other fields, we shouldn't use AI to get rid of our responsibilities - see the sketch shown in the lead image - we should use it to strengthen responsibilities and accountability.
Subscribe to this newsletter and follow me on social media for more publications and thoughts:
LinkedIn
Mastodon