VIDEO — 26 May 2026

Algorithms, personal data, AI in warfare and curiosity about artificial intelligence

A studio interview on Radio Antena M about algorithms and our data, AI slop, critical thinking, AI in warfare and using AI in everyday life.

Published in

Radio Antena M

Role

Guest

Watch on YouTube

In the Radio Antena M studio we talked about how vulnerable society is in the age of algorithms. We covered our personal data, AI-generated content and disinformation, AI in warfare and in the fight against corruption, and how each of us can use it in everyday life.

Algorithms already steer many of our decisions

Technology is advancing faster than ever, and artificial intelligence has really stepped on the accelerator. Many of our decisions are now directed by algorithms, because we have allowed it. Our personal data is used to tune them to serve us products and steer our way of life, from what we buy and watch to how we behave in a relationship, in marriage and as parents.

With free services we are the product, and for harmless fun, such as a digital avatar made from a photo, we hand over very sensitive data. The platforms get to know us better than the people closest to us, and we are so predictable that an algorithm with very little information can tell which way our thinking will go, which is why it can feel as if the phone is listening when it shows us a particular ad.

The European Union has only just begun to regulate big IT corporations such as Google and Facebook through the EU AI Act and other rules, and these companies, with unlimited resources, would rather pay expensive fines than stop collecting data on a mass scale.

AI slop, copyright and slow regulation

I am optimistic that institutions will manage to protect each individual, although the current situation is catastrophic. The social networks themselves have recognised that they are flooded with AI slop, very low-quality AI-generated content that holds a few seconds of your attention and carries no real value.

People have started to push back, and OpenAI shut down its experimental video app in which 100% of the content was made by AI. I enjoyed a funk cover of Linkin Park today, yet I understand the artists who make such versions with a lot of work and talent, while someone made this one in minutes with a model that costs 20 euros.

Social networks now label generated content and give it less weight in the algorithms. Regulation will also have to sort out copyright, a grey area for now: OpenAI and Anthropic are in court for using music, books and photographs without asking, yet their models are already trained. I hope we will reach an acceptable model within 5–10 years, although for regulation at a global level even that is an ambitious estimate.

Critical thinking is the only protection for now

The narrative around AI is extremely negative. I have dealt with deepfakes myself, and I don’t think we can ever talk too much about how serious that problem is. Until regulation takes hold and the platforms help us, critical thinking is the only protection an individual has.

Before you share

01When a sensationalist story or photo stirs up anger or disbelief and you want to share it straight away, stop for a second.
02Ask yourself whether it is real, and spend another half a minute or a minute checking the sources.
03Ask ChatGPT or another AI tool to check whether credible sources have reported the story, because it does excellent research and answers quickly.

That way you can be your own investigative journalist and avoid damaging your reputation by sharing unverified content.

Drone swarms and autonomous targeting

In warfare, AI is currently used mostly to cut the cost of targeting, because missiles are expensive and so is putting soldiers in the field. Nuclear weapons changed far more, and I think the Cold War was the peak of the risk of global self-destruction. In the war between Russia and Ukraine we can see AI and robotics meet in a slightly creepy Terminator moment: swarms of cheap, small drones carrying large explosives fight each other and hit very small targets from far away, and AI helps them arrange themselves and perhaps avoid anti-drone weapons.

The public only learned that autonomous targeting exists when Anthropic refused it, along with the mass surveillance of American citizens, in a US Department of War tender to bring AI models into the Pentagon’s warfare systems. Anthropic got into serious trouble over it because, as far as I know, OpenAI and Google agreed. Israel has an AI-based system called The Gospel that uses autonomous targeting and has been used in Gaza, where so many innocent people were killed, and it is an open question how much of the targeting AI did on its own and whether a human in the loop approved each attack.

Corruption, public procurement and the human factor

To fight corruption in public procurement, we first need the will, and it is missing because every new government finds it convenient to keep a little room for corruption, which is a global problem and a human phenomenon. We have also skipped many basic controls, such as several levels of decision-making and checks on every procurement before it is published, which paper and pencil can handle without any AI.

AI is better suited to optimising and speeding up these processes. If I know that AI is scoring me, I can tune my documents to become the perfect applicant, and AI cannot walk over to my office to check whether I am a phantom company. That is why I would keep people as the final decision point, because otherwise we would hand the helm to a system whose workings under the bonnet we still don’t understand.

How AI can help with learning, health and work

Fear is justified if we focus only on the negative consequences, but AI is a tool, like a stone, which can be used to build a house or to hurt someone. Install one of the apps, play with it and see how it reacts.

—Learning. It can make you a training plan, help you learn a topic in the way that suits you and test you before an exam, and it does this superbly.
—Health. Ask it for advice when you first notice symptoms, and always make the doctor the last step. It is very good at spotting the first signs of a serious problem, and 80% accuracy beats having no idea what to do.
—Work. If you fear that the processes filling perhaps 80% of your working day will be automated, let the change start with you: offer to lead the automation of that work and put your hand up for the tasks AI cannot finish.

Test the tools in your own work

Be extremely curious about artificial intelligence and try it in your own work, whether you are a car mechanic, an electrician, a financial expert, a doctor or a politician. Talk to it by voice and use the live camera. When my printer broke, I turned on the camera and talked to Gemini for a few minutes, and it recognised the model and told me the first three things that would probably fix it, so I solved the problem in minutes.

ChatGPT is like Viber here, the app everyone got used to first, yet Claude from Anthropic and Gemini from Google have turned out to be a little better, and you can test all three for free. A 20-euro subscription is worth it if you use it often, and AI has saved me so much time on documentation, emails and tasks it completes on its own that I can be there more for my employees, my child and my family.

More conversations & talks