Does democracy need AI?

Democracy is, theoretically at least, a citizen-centric system of government. In practice, democracy is a spectrum of governance models that are gamed to varying degrees to benefit the powerful. The trend appears to be towards oligarchy and this manifests itself in the poor getting poorer.

Potentially AI may have a role to play in reclaiming democracy.

The future of government

Government’s relationship with AI is likely to evolve in the following stages:

• AI-enabled government, eg. smart cities

• AI-driven government, eg. predictive policing.

• AI government, ie. all policy decisions are digitally reached.

The idea of being governed by software seems strange, but it is already happening in respect of taxi and delivery drivers. So having a software boss is nothing new.

Much wants more - The problem with humans

Democracies are typically vulnerable to politicians who are power hungry and most likely sociopaths. Their allegiance stack comprises:

1. Self

2. Party

3. Country

4. Planet.

In some cases, the interests of their place of education rank in the top two.

Then there are the powerful who can shape policy by levering their influence / party contributions to get what they want. This can range from favourable taxation and imposing tariffs on foreign competitors to having the local police dismantle peaceful protests that threaten their commercial interests. These are the puppeteers. Our elected leaders are the puppets.

Humans are naturally greedy. Back on the savanna we would take more than we needed because we did not know when we would eat again. Capitalism has provided a model for human greediness to happen at scale. Thus a few benefit at a cost to the many.

A democracy that gave the average citizen an equal voice would go a long way to addressing this twisted form of government. In theory, AI could be configured to genuinely have the best interests of all stakeholders, including the planet, at heart.

“Skynet, how can I help you?”

The benefits of AI to society are well established, healthcare, disaster recovery, education and in environmental conservation. But we must be mindful of the downsides. Genetics is a similar case. The benefits are clear, but the associated risks are not yet fully understood. We must progress carefully.

AI is only as good as the data that feeds it. Most organisations of any size typically have a data cesspit rather than a pristine data lake. Also AI can mix up causation and correlation and thus biases emerge.

Whilst it is possible to programme values into AI, the emerging principles and moral code may well deviate from what is intended. AI’s lack of contextual understanding might cause it to lean towards fundamentalism.

Trickle up economics

Naturally, not all citizens trust their governments and thus have disengaged from political life. Trust is key. Democracy only works with an informed and engaged citizenry. Disengagement, on the other hand, works well for oligarchs.

If needed, citizens can be distracted by fear (“the enemy is near”) or sedation (cheap Internet and alcohol).

Humans respond well to having a sense of belonging, being on a path to mastery and enjoying some degree of autonomy. Liberalism addresses mastery and autonomy. However it presumes we are all self-interested lizards, when in fact sociality is what keeps us alive. It also provides the framework to ensure the rich become richer.

It is not a crime to be rich and its not even immoral. But sometimes it can lead to behaviour that appears stupid. Smart billionaires must surely think of the planet and societies they are leaving for their descendants?

Again AI has a role to play in ensuring that all stakeholders are treated fairly. Everyone wins when we have an economically prosperous and inclusive society. Think of it as trickle up economics.

AI offers a more equitable means to ensure the interests of all parties are met. It is likely that citizens will eventually have their own digital AI-driven avatars that will handle their civic responsibilities in respect of influencing government. So rather than being asked to vote every now and then, citizens will contribute to dozens of policy decisions every day. Even though government will be at least partially software-driven, it will behave like a living, sensing organism that is acting in the best interests of its stakeholders.

We need to talk about AI and government

Our choice is between naturally flawed humans and an untested AI. The latter is creeping into society unimpeded. AI would work nicely for dictators. Think drunk driver on a semi-tested self-driving car.

Democracy seems like a reasonable way to proceed. Despite millennia of human-led democracy, it still feels wired for unfairness. Perhaps we should give AI a shot? But perhaps we need to remind ourselves that we are social animals. It is better if we move forward as a group.