ARTIFICIAL INTELLIGENCE AND POLITICS: Who controls the algorithms?

Algorithms increasingly determine our lives. From social networks that influence our media consumption behaviour to AI systems that can steer political decisions. Artificial Intelligence (AI) is developing into an ever more powerful actor. But who controls the algorithms that help shape our political and societal realities?

AI in political communication

AI has revolutionised the way political messages are disseminated. Election campaigns use data-driven approaches to tailor content specifically to voter groups. Algorithms personalise these and thereby increase the effectiveness of political advertising. However, problematic dynamics emerge: filter bubbles and echo chambers.

Algorithms show users preferred content that corresponds to their existing beliefs. This mechanism can lead to opposing opinions being filtered out and polarisation being reinforced. According to a study by Lazovich (2023), on platforms such as ChatGPT, algorithms adapt content based on a user's political affiliation.

Search engines can also influence opinion formation. Epstein and Robertson (2018) demonstrated that manipulating the order of political search results can change users' preferences. Up to 20% of participants in an experiment in this context changed their opinion in favour of a party when the search results were prioritised accordingly. This shows: algorithms can not only reflect political opinions, but actively help shape them!

AI in the political decision-making process

AI is not only used in political communication, but also directly in decision-making processes. Governments use AI systems for forecasting, for example to assess economic risks, or for migration control.

One example is the use of predictive policing, in which algorithms are supposed to predict crime. However, these technologies are opaque and error-prone. Predictive policing models are based on historical data that often reflects prejudices and existing discrimination. This leads to disadvantaged groups being disproportionately targeted by law enforcement.

Fake news and disinformation

Another growing problem is the spread of disinformation. Deepfake technologies and automated bots use AI to disseminate false information credibly and on a large scale. This became evident during the 2016 US election, when social bots demonstrably influenced public opinion (cf. Woolley and Howard (2018)). Similar campaigns were observed during the Indian parliamentary elections in 2019.

Who controls the algorithms?

The power of algorithms raises a central question: who controls how AI is used? Currently, large technology companies such as Google, Meta and OpenAI dominate the market. Although political regulation, such as the EU Digital Services Act (DSA), is intended to promote transparency and accountability, it is often insufficient. It remains unclear how global companies can be held accountable.

Scholars such as Florini (2018) emphasise that national regulation is not enough to control the global impact of AI. International cooperation is said to be crucial in holding technology companies accountable. One proposal involves the creation of independent international organisations that monitor and regulate the use of AI and thus also algorithms.

Democratic control of AI is also crucial. Binns (2017) calls for greater public involvement to ensure that AI systems are in line with ethical and democratic principles. Technical transparency alone is not enough – public debate, education and political engagement are needed to regain control over algorithms.

What can the individual do to be controlled less?

In addition to regulation by politics and international organisations, each of us can assert our influence to be less strongly controlled by AI.

The following points are merely intended as ideas and food for thought to reconsider personal options for action with regard to AI.

• Strengthen media literacy: critically question why certain content is displayed.
• Choose platforms more consciously: use search engines such as DuckDuckGo that guarantee transparency about algorithms.
• Protect data: consciously minimise personal information.
• Avoid filter bubbles: manually search for content and prefer non-personalised offerings.
• Become politically active: advocate for stronger regulation and transparency.
• Question AI critically: do not accept results and recommendations from AI systems without scrutiny.

Conclusion: The power of algorithms

Artificial Intelligence holds enormous potential to transform politics and society. However, without clear rules and control mechanisms, there is a risk that AI will become a tool that undermines democratic values. Control over algorithms should not rest solely with technology companies. Instead, there is a need for broad public debate, effective political measures and international cooperation. Humans should not be seen as victims of the system or technology, but as autonomous, decision-making and acting individuals.

written by Nele Jahrsdörfer

Literature:

Woolley, S. C./Howard, P. N.: Computational Propaganda. Political Parties, Politicians, and Political Manipulation on Social Media, 2018.

Binns, Reuben: Fairness in Machine Learning. Lessons from Political Philosophy, Conference on Fairness, Accountability, and Transparency 2017.

Lazovich, Tomo: Filter bubbles and affective polarization in user-personalized large language model outputs, Boston 2023.

Epstein, Robert/Robertson, Ronald E.: The search engine manipulation effect (SEME) and its possible impact on the outcome of elections, in: Proceedings of the National Academy of Sciences 112, 33 (2018) pp. E4512–E4521.

Floridi, Luciano et al.: AI4People – An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles and Recommendations, in: Minds and Machines 28, 4 (2018) pp. 689–707.