Ethical Pictures in the Age of AI – An Opinion Piece

Kirchschläger_Peter_G by Prof. Dr. Peter G. Kirchschlaeger, November 2025

Ruminating Cows

So-called generative «artificial intelligence (AI)» – more adequately coined “data-based systems (DS)” because they lack intelligence – can produce pictures as we wish. They do so at the price of violating the human rights to data protection and privacy, infringing copyrights, and causing major negative ecological impact due to massive energy and water consumption for cooling down the data centers behind generative DS.

Generative DS deliver pictures on demand – like a ruminating cow: humans feed the cow with data, the cow digests the data and spits out something. Besides the above-mentioned fundamental normative problems, nothing genuinely new comes out of this cow. It is not innovative. It remains captured in the past and in already existing data.

Comuzi / https://betterimagesofai.org / © BBC / https://creativecommons.org/licenses/by/4.0/

Bias and Deception

In addition to the ethically relevant transformative impact this technology-based value-creation process has on human labor, the data the cow is fed with is never fair, neutral, or objective. The digestion of the cow – in other words, the algorithms – is never fair, neutral, or objective either. Among other issues, biases and stereotypes, which can lead to or foster discrimination, come with pictures provided by DS.

These pictures may have little or nothing to do with reality. Therefore, if a company or an NGO tries to attract clients or supporters with pictures generated by DS, it avoids or destroys a relationship of trust because these pictures do not represent the reality a company wants to cultivate economically, or an NGO strives to gain support for in overcoming a challenge. Ultimately, this consists in a deception which is poison for trust-building.

The criterion of truth does not play any role at all for DS. DS do not care about what they produce. They just “do” what they were trained “to do”. It is not even relevant to them what they “do”.

While acknowledging the cost-efficiency of pictures generated by DS, what sense does it make to deceive each other with DS-generated pictures of a non-existing “reality” while we still live – as humans, as global citizens, and as consumers – in the real world? There will not be any sustainable impact – neither economically, ecologically, nor socially – from pictures generated by DS. Lies have short legs …

"Humans – not machines – can be trusted because trust needs an interpersonal relationship as a basis."

Prof. Dr. Peter G. Kirchschlaeger

Made by Human Hands

As long as DS embody the ethical problems mentioned above, a picture, in order to be considered ethically legitimate, should be made by human hands. Humans – not machines – can be trusted because trust needs an interpersonal relationship as a basis. This is not possible with DS.

In addition, humans are able to reflect upon and relate critically, from an ethical standpoint, to the process of taking and using a picture.

Furthermore, contexts and humans which are on the pictures should – as a condition sine qua non – have a say in the taking of a picture. Informed consent is a necessary condition for taking as well as showing a picture. This participatory approach serves the respect of human dignity, freedom, and autonomy of all humans.

Beyond that, the creation or the enhancement of biases and stereotypes through pictures should be avoided.

Finally, if pictures are considered as parts of a communicative process, then in order to avoid one-way communication between a fake DS-generated picture and humans, there is a need for human-made pictures.

Download AI White Paper and Checklist

Fairpicture has developed a white paper and checklist on how to deal with AI-generated images from a practical perspective. Download it now.

Download