Until recently, when most of us needed information, we’d turn to search engines, particularly Google. “Google it” became virtually synonymous with the act of searching itself.
Now, we increasingly hear people say “I’ll check ChatGPT” – referring to an Artificial Intelligence (AI) system. In recent years, AI has become increasingly embedded in our daily lives. The same is happening in politics and elections – in short, in the democratic process.
“There are both risks and opportunities”, said Kristian Bartholin, Secretary General of the Council of Europe’s Committee on Artificial Intelligence, speaking to philenews during his recent visit to Cyprus.
He was here for an AI event organised by the Parliamentary Institute for Parliamentary Democracy and Participatory Democracy of the House of Representatives.
As he explained, AI is a powerful tool that can help make societies more participatory and inclusive.
While AI’s capabilities for message dissemination are remarkable, this same power can become dangerous due to the rapid production and spread of fake news.
This is precisely what concerns Bartholin – that it could turn against us.
“If it’s used to broaden democratic participation, I think that’s excellent. However, if it’s used to control people, often in subtle ways, then it becomes problematic,” he explained.
He believes the greatest danger lies in how difficult it is to distinguish AI-generated content from reality.
“AI doesn’t give correct answers – it’s ultimately algorithms predicting the most likely correct answer among many possibilities. Sometimes it might seem like we’re interacting with a human, but of course, we’re not”, he emphasised.
AI has become firmly embedded in our lives recently. Do you see this as a positive or negative development?
I believe it’s a positive development, and AI will likely prove to be a technology that genuinely benefits humanity. But like all technologies, it has its downsides, and I think we shouldn’t either underestimate or exaggerate these – we should maintain a healthy scepticism about its use.
We shouldn’t become overly dependent on it or rely on it entirely. We must maintain confidence in human intelligence, which has served us well for thousands of years. Even the term ‘Artificial Intelligence’ is somewhat misleading.
In reality, we’re dealing with algorithms that predict the most likely correct answer among many possibilities. While it might sometimes feel like we’re interacting with a human, we’re not – it doesn’t have the same holistic approach as a human being.
I think we should be optimistic and embrace it in many low-risk areas like science, education, healthcare, and for tasks that would otherwise require significant effort. We can’t replace human intelligence with artificial intelligence, and we must ensure we maintain human dignity and autonomy in its use.
Let’s say you ask an AI system to compare candidates’ positions in an election and suggest who to vote for. Would you take its answer into consideration?
No, I wouldn’t. I believe that when we vote, make friends, or choose colleagues, we make those choices ourselves. It’s a deeply human decision made by humans.
Of course, we read and stay informed through media about what different candidates stand for, and that helps us make decisions. There’s nothing problematic about that, since we know from the start a media outlet’s political leanings – whether it’s a right-wing or left-wing newspaper, it has a specific viewpoint that we take into account.
But with AI, there’s a tendency to accept its answer as correct without question, perhaps because as humans we tend to rely on machines and believe they make better choices than we do.
How do you see AI’s role in participatory democracy in the coming years? Does its use present more advantages or disadvantages?
Well, it entirely depends on how it’s used. If it’s used to expand participation in democracy, I think that’s excellent. On the other hand, if it’s used to control people, often in ways that aren’t immediately apparent, then it’s negative.
For example, we have AI systems that can recognise people and correlate them with databases containing information like their tax payments and where they live.
Imagine all this being connected to CCTV systems. Every action we take would be recorded.
Then, imagine someone participating in a protest and being immediately identified as having unpaid taxes. “Oh, this person is protesting but owes us money – let’s find them to pay their taxes and a fine,” some in government might think.
And that’s just the mild side. In a real dictatorship, people could end up in prison, face punishment, or their families could face consequences.
That’s why I believe if AI is used for surveillance, it could potentially have a very suppressive effect on participatory democracy, whether in real life or online. That’s why I’m concerned that using AI for surveillance purposes could suppress democracy.
We’re concerned about misleading use and manipulation
Let’s talk about proper usage then. How can AI promote democracy? Can you give us some successful examples?
That’s a good question about AI implementation because we’re still in its early stages of development. Some experiments worldwide show it can be used to increase participation, not just in electoral processes but in other social processes.
Good examples include when it’s used to help people better express their views, or when it helps us translate text to understand a political position or party. There are concerns, of course, mainly regarding the production of deepfake stories and images or anything else.
The misleading nature of AI is a serious issue. For example, a fake video of a politician saying strange things. We hear their voice, see their image, all created by AI.
But in the end, it doesn’t seem so strange because we can associate it with a specific person.
If this becomes the norm, we’ll start losing our trust in reality. And once that trust is lost, society’s smooth functioning will stop. And we know very well what that could mean.
But this is already happening. In recent elections in India, we’ve seen AI-generated videos being widely used.
So far, we’re used to political advertisements. What we’re not used to is AI use, and we don’t approach it critically. We don’t know how it works yet.
This will come with time. I mean, if we go back 100 years, squares were filled with people attending election rallies, and they didn’t trust radio or television. In the future, algorithms might help politicians reach even more people.
I don’t see anything wrong with that, but we need to be sure that politicians are actually saying what we hear them saying.
So how can we ensure that the risks from AI use in our daily lives won’t harm democracy?
I think there need to be some basic, let’s say, protective guidelines put in place. Perhaps some particularly misleading uses of AI should be banned, or we need to ensure there’s clear labelling when something is AI-generated.
The European Union is working towards this, as is the Council of Europe. And as I’ve already said, what particularly concerns us is AI’s ability to be used in misleading ways and to manipulate.
To some degree, there’s always manipulation in politics – politicians promise things and don’t deliver. What’s frightening is that we can’t separate what’s real from what isn’t.
Moreover, even if we’re told something isn’t real but AI-generated, we tend to forget the difference if it seems plausible and aligns with what we know.
Strong democratic tradition in Cyprus
Considering Cyprus’s unique political landscape, what challenges and opportunities do you see AI creating for participatory democracy?
Cyprus has a proud democratic tradition. You work hard at it, and your system functions quite well. AI can play a positive role if you use it to increase and improve citizen participation in public affairs.
I can’t give specific examples, but algorithms can be used flexibly to respond to various needs. In any case, you need to see what helps people and how the situation can be improved. Find out what interests them so they participate in political decision-making processes.
If AI can help with this, that’s good. But as I said, digital literacy is required to use it properly and prevent it from becoming a tool for manipulation.
Do you believe AI will change democracy and politics in the future? And if so, how?
I hope it changes them for the better, increasing citizen participation in democratic processes. However, my fear is that it might lead to greater political polarisation because we’ll lose what keeps societies united – trust – and we’ll become even more divided.
We won’t trust anything because we won’t know what’s AI and what isn’t, and we’ll end up losing our bearings. It produces easy answers, easy solutions that we trust without hesitation.
And if we rely on AI for political decisions, we might walk down very dangerous paths. Today, we all know not to trust everything we find on Wikipedia, but at the same time, it’s an easy solution.
If I need to check something, I myself will look at Wikipedia to find information about, for example, a politician I don’t know in another country. But can I trust this information? And the danger will become even greater if we let AI choose for us what we’ll hear and what we’ll read.
From your experience at the Council of Europe, what lessons have we learned so far from AI implementation? What steps will be taken to ensure safe use of this technology?
Our experience so far is that every technology from antiquity to today has both positive and negative aspects. Technology has no conscience; it’s people who make choices.
My colleagues and I work on how to involve AI in political decision-making processes to make democracy, as we’ve already said, more participatory. We need rules. We need protection from direct manipulation.
It can’t be used to suppress political speech and prevent people from participating in political processes because they fear authorities know about them or potential punitive measures against them. Perhaps it’s similar to the printing press in some ways.
Its invention gave people the opportunity to print and spread a message everywhere. For better or worse, the same thing can happen now.
Measures are being taken at the European level. We need to strengthen discussions in our societies, between politicians and citizens, about how we want and don’t want to use AI. The technology is here, and it’s up to us to make the choice.
Digital literacy is key
Could education play a role, with special courses in schools?
You’re right about education’s role. I think digital literacy, and specifically AI literacy if you will, is key.
If we know what we’re dealing with, if we can understand it, then a large part of what could go wrong is eliminated.
I believe it’s very important for children to learn about AI early and be trained to distinguish what’s real and what isn’t. They should be taught that AI can’t replace their own critical thinking and opinion but that it’s a tool to help them complete certain tasks.
However, I’ll tell you that I’m more concerned about older people than children. The elderly are the ones who really struggle to understand what AI is and what it can do.
