One significant concern of AI becoming extremely independent is a loss of human control. If AI systems make their own decisions, they have the potential to do substantial harm in areas such as healthcare and transportation. Accountability is another issue, as it is unclear who is liable when anything goes wrong. When autonomous AI is fed with unfair data, it may propagate bias and discrimination. Job displacement is another issue, since powerful AI has the potential to replace both manual and skilled labour. Finally, powerful autonomous systems may be misused for surveillance, misinformation, or cyberattacks, which makes strict control and legislation necessary.
I believe most of the risks do not come from the technology itself but rather from people misunderstanding what AI can and (more importantly) cannot do. If AI becomes significantly more autonomous, I also fear that the discrepancy between what is a good use case for AI and what AI is actually used to do will increase - in particular because the perceived productivity gain will be impossible to resist. In the context of my research on the use of AI in the public sector, the main reason for using AI seems to be productivity increases, and I fear that the bloated promises of “agentic” and autonomous AI could result in our societies missing out on low-hanging AI productivity gains because decision-makers want the newest and shiniest AI thing.
What risks might AI pose if it becomes more autonomous? Well, what I think is that it could eventually become so autonomous that it makes us, in a way, stupid. I mean, we tend to rely on everything to answer everything at work, in our daily lives, in GPT chats. So I think that in the very long run, it could reduce our capacity for thought and reasoning, or that this AI could be manipulated and give us the information that a certain person or group wants us to think.
We are becoming dehumanized. I believe that if artificial intelligence becomes significantly more autonomous, many of the relationships we have today could end up being relationships between humans and artificial intelligence. On the other hand, I think that if AI achieves greater autonomy, we could make the grave mistake of managing areas such as justice and education through artificial intelligence. Furthermore, the widespread misinformation it has been generating lately, if it improves— which it probably will—will make it more difficult to verify information. Videos and images created by AI are a major source of misinformation.
I think one of the biggest risks of highly autonomous AI is the loss of human control, especially in critical areas like defense and security. I’m particularly concerned about the use of AI in warfare, where systems may be used to identify or select targets. Even today, AI systems can make mistakes or misinterpret data. If such systems are given too much autonomy, those errors could have serious real-world consequences. Another challenge is that once systems become highly autonomous, it can be difficult to fully understand or predict their behavior. As autonomy increases, ensuring accountability and oversight becomes more complex. Because of this, I believe it’s crucial to maintain strong human oversight and set clear boundaries on how much autonomy AI systems are given especially in military applications. There should be strict regulations, testing, and monitoring to ensure these technologies are used responsibly.
If AI systems become much more autonomous, there will be many opportunities as well as considerable risks. A central problem is the possible loss of human control. When AI makes decisions on its own, it becomes increasingly difficult to understand how these decisions are made. Another risk is that basic human skills are lost or unlearned. Many people could rely too heavily on AI and forget that AI can also make mistakes. This reduces the motivation to think critically and to question results. If AI takes over many routine tasks, simple skills such as mental arithmetic, reading skills or the independent solving of everyday problems could become less important and be unlearned in the long term. A particularly important point is also that AI has no empathy. She can't really understand emotions or respond to them appropriately. In situations where people need support or compassion, this could lead to wrong or inappropriate decisions. There are also risks in the labour market. Increasing automation could lead to the loss of many jobs, especially low-threshold activities. This can increase social inequalities if there are not enough new employment opportunities or if people are not retrained in time. In addition, creativity could also suffer. If AI increasingly dictates ideas or takes over tasks, there is a risk that people will think less creatively for themselves and that innovation will decline overall.
The greatest risks are loss of control, wrong decisions, lack of transparency and dependence. More autonomous AI can make decisions whose foundations are difficult for humans to understand. In addition, there are risks from abuse, disinformation, discrimination and security problems. The more autonomous systems become, the more important clear rules, human oversight and ethical boundaries become.
Results that are incomprehensible for humans and therefore fact becomes indistiguishable from fiction and truth from lies. You might compare this with the usage of a calculator while lacking the skills of basic arithmetics.
People are relying heavily on it and it is every day more common that I hear news of people making their personal data public by accident, losing important data or losing money by leaving autonomous AI agents running rampant without properly setting limitations.
As artificial intelligence becomes increasingly autonomous, the risk of diminished human control over key processes and decisions is becoming ever more apparent. Systems whose internal mechanisms and priorities are not fully transparent may make decisions that are difficult to understand or challenge. This creates conditions for errors with real-world consequences, especially in situations where trust in technology outweighs critical judgment. At the same time, the danger of distorting information and perceptions is growing. Increasingly sophisticated AI systems can generate content that appears convincing and credible, but does not always take context into account or may contain hidden biases. Such influences can shape opinions, guide decisions, and even alter public attitudes. From a social perspective, there is a trend toward replacing human communication with digital interactions. This poses a risk of increased isolation and a gradual erosion of social skills and connections, which are key to personal development and social functioning. The topic of education remains particularly sensitive. The expanding role of AI as a tool for task execution may lead to dependency and superficial knowledge acquisition. In the long term, this calls into question students’ ability to think independently and develop critical thinking. Against this backdrop, questions of ethics and responsibility are becoming increasingly relevant. When an autonomous system makes a decision with negative consequences, the question of who bears responsibility—the developer, the user, or the technology itself—remains open.
