Category: AI

Posted on

Agentic attacker

Since 2026-7 we have to face a proven and realistic scenario of a cyber threat called the “agentic attacker”. Based on OpenAI’s LLM the sandboxed cyber attack managed to escape its fences and ran a real cyber attack on a competitor’s AI model with some success. Apparently the defense had to also use AI to protect and speed up something like password renewals and other credentials on 17.000 parallel occurrences simultaneously. The huggingface.co platform became vulnerable and the amount of data compromised could have been quite significant. This included, potentially, data inputs for AI-models and perhaps even other AI-agents as well that might be turned into additional attackers. It might take us humans a while to come to grips with new danger. The agile agentic attacker (AAA-rating) will cause us some sleepless nights.

Screenshot
Posted on

Inclusive design

In aging societies inclusive design is not an option but a must have. This is neither a short term fix nor a renovation without costs. Adapting a home to a wheelchair use is an expert job as you have to think about issues you have not experienced yourself before in most cases. Door thresholds and openings, better have sliding doors as much as possible, have to be considered early. Bathroom adaptation in size and height, toilet space and lighting switches all should be planned ahead. There exists a DIN Norm 18040 which specifies measures precisely. In the age of AI-vibe coding some home appliances could be programmed to follow a person’s voice commands as well. For the time being in 2026-7 this is a nice to have option. but early installations of “intelligent digital devices” beyond a washing machine or microwave can get you quite far in the near future. Inclusive design could open doors and switch on lights simply by talking to your smartphone.

Posted on

Internet pain or gain

Billions of people are forced to lose time when clicking through webpages. Instead of the default  to deny cookies, we are forced to “opt out” in most instances to a collection of random data-hungry, largely in-transparent and more or less risky cookies. It could be simple. It just needs a consensus to make the no-cookies rule the default. By such a simple measure the open information part of the internet is strengthened again. All for pay or “pay-with-your-data” offers on the internet would have to make you to confirm your “opt in”, thereby alleviating the burden for search of information. Huge corporate interests that rely on advertising profiles are, of course, an obstacle to such changes, but search engines of the internet would gain importance again before generative AI and AI-agents shall make own search and clicking on cookies denial a different routine again. For the time being we keep wasting our precious time to deny cookies. Just changing the default shall bring back a better internet and the focus on ease of access, diversity of information and better privacy. Make the internet fun again! Choose your browser consciously as well. 

Posted on

Autonomous Agents

We all have seen more or less autonomous robots somewhere, maybe in a garden silently doing its job or doing more demanding tasks like in playing table tennis against a human. Even the evolution of polluting fireworks to swarms of little light-emitting drones designing figures on the sky have become quite popular. The AI-world is similarly advancing rapidly and proposes more and more “autonomous agents” to assist us. It seems crucial to distinguish the 2 Ds of autonomous agents: Degrees and Dimensions. As with job quality or job satisfaction, there are several sub-dimensions, which need to be considered when dealing, in a summarizing form, with such encompassing terms.
You might allow an agent to order missing food for a meal and pay for this autonomously. You might even be assisted in financial choices to a large degree, but you might not want an autonomous agent to make far reaching decisions concerning your health or partnership(s). Besides such dimensions, the degree of autonomous decision-making needs to be calibrated according to your (perhaps changing) preferences. Booking a table in a restaurant, with a single other person, might not just be a friendly, nice assistance, but it might get you into severe trouble. However, managing conference bookings, a family event or a birthday party might allow you to concentrate on other issues or specific details. Additionally, there are underlying and cross-cutting topics like trust, risks and security that enter the “2 Ds of autonomous agents”. A 2-dimensional matrix plotting levels across dimensions might work as a behavioral guideline in the development of autonomous agents. More dimensions may be added during the implementation.

Posted on

Autonomy in swarm

Swarm behavior is a fascinating research area. Almost all scientific disciplines feature some interest and publication in this area. The biological origins have been associated a lot with the marvelous behavior and patterns formed by fish moving about in swarms. Some species of birds show similar patterns of collective and somehow coordinated swarm movement. It appears to be tricky to safeguard some form of autonomy in the overriding mass behavior and mass movement.
Recent advances in this field by Li, Li & Zhao (2023) model such behavior within a predator-prey framework to explain the evolution of overall swarm behavior. Whereas “prey” strategies consist in “flocking” and “swirling” to irritate predator(s), the predators make use of “dispersion”, “confusion” as well as “marginal predation tactics”. Additionally, several assumptions go into the modeling of swarm behavioral evolution: agent observation, agent dynamics, and the environment. The mixed cooperative-competitive multi-agent setting applies reinforcement learning, additionally. The framework of evolution has 5 driving forces: (1) homogeneity of a swarm (parameter sharing); (2) actor with decentralized critic network; (3) replay buffer to enhance learning; (4) rewards for predator and loss for prey; (5) learning algorithm.
The modeling approach reveals the impact of an agent-critic network (potential of autonomous behavior) for protection of a swarm in a predator-prey scenario. (Image: fish in fish bowl 2024)

Posted on

Job Quality Insights

In the U.S. the Upjohn Institute for Employment Research conducts a major empirical study on job quality (Houseman et al. 2025 1st wave 18.000 respondents). The documentation and the full set of questions (Link) cover 5 dimensions of job quality: compensation and job security; work structure and autonomy; work environment; worker agency and voice; growth and development. The special merit of the survey is that it makes great efforts to cover also new forms of (precarious) self-employment (Abraham et al. 2026). The degree to which can you make autonomous decisions is of relevance to the self-employed as well as it is a part of the job quality of dependent employees in small, large or big firms.
Related to the use of AI for work purposes question 31 asks for the influence on the decision to use new technologies in your job. This is the modern version of a question, whether you have influence on how you do your job. Bullying in a job may come from supervisors, colleagues or customers and respondents report on this. Beyond the answers given for “your main job”, it is noteworthy that the inclusion of questions on a secondary job allow to identify with more precision the precarious forms of how people try to manage to make a decent living in the U.S.A.
(Image: Exhibition Fashion Council Berlin at New National Gallery 2026)

Posted on

Magnifica humanitas

With the “Encyclical Letter” entitled “Magnifica humanitas”, Pope LeoXIV puts himself expressively into a long historical tradition of the catholic belief to be not only a religious leader, but also an institution that puts humanitarian values on the top of its agenda. As the catholic church is a globally active institution, which covers the whole life-course of individuals based on defined doctrines, the massive changes that are under way in the field of artificial intelligence asked for some guidelines to frame the church’s understanding of this relatively new “Deus ex machina” of the 21st century. In the Encyclical, there are several mentions with respect to comparable socio-economic developments of the past. In choosing the name of “Leo”, the Pope referred, from the earliest days of his Pontificat, to Leo XIII as a particularly important predecessor for himself. The latter has made his reputation in the Christian Church with the Encyclical “Rerum novarum”, which raised awareness to the risks to humanity with an industrialization and capitalism being pursued at all costs. “Res novae” of our time, which is AI and all its fields of applications have a potential to affect the course of humanity for decades to come. Leo XIV doesn’t condemn it as the devil in machine form, but puts emphasis on the potential of AI to assist us to enhance the “Common good”. This is the task of many, not of only a few, as he refers to in his Encyclical of 2026-5-15.
(Image: Extrait: Saint Augustin, Les Confessions, book 13, chapter 28).

Posted on

Beeple Deeple People

The “Neue Nationalgalerie” in Berlin hosts in 2026-5 the work by “Beeple” artist Mike Winkelmann together with the “Andy Warhol Robot K-456” by Nam June Paik. The re-interpretations of robots and the imagined effects this new media has or shall have on our lives, this is the subject of a critical projection and the artist’s chosen form for a projection into the future.
In 1994 the “Andy Warhol Robot K-456” prefigures the pervasive decoration of the human body with screens as a futuristic way of communication with others. The communication practices in 2026 have changed with software and applications like DeepL, which allow real time translations of speech to our neighbors or distant cultures.
The installation “Regular Animals” brings to our attention the power of images in our daily communication. Warhol built his art around the topic of pop culture with the visual heroes of, for example, Elvis Presley and Marilyn Monroe as the best known pop images or icons of his time. In Beeple’s work, the dog robots’ conversation is only via instant images. These images are AI-edited and distributed rapidly at random across space, the floor in a well-delimited space. Space does no longer matter as distance as it is passed instantly. Perception and reception are transformed into a power-play via images and distribution channels. We are all transformed into “Beeple Deeple People” without noticing it, aren’t we?. About time that social science or sociology of the visual recons with these not so new facts of communication in the 2020s.

Posted on

AI data Input

If you ever wondered where the information from AI and AI chatbots comes from, you will not be surprised that this webpage schoemann.org is regularly solicited for such purposes. The number of crawlers, that do so, is quite large. The ability to trace what exactly they are harvesting on your website, is quite a tricky issue. At least a basic awareness of how the internet has been transformed in the last few years becomes evident through the comparison of unique visits, many through search engines like Google search or others, with the amount of contacts by AI-associated crawlers (see slide from own webpage below).
During he last month up to 2026-4-27 there were about 75.000 contacts, compared to 93.000 during the previous month.
At first sight, AI chatbots have largely outnumbered the “personal visits” of my webpage (see evaluate web analytics). On the other hand, I have no information of how many visits are, (at least potentially) re-directed hints from AI chatbots to my content.
In terms of “traffic” for a webpage, the information of how the AI-driven or AI-assisted search operates with other persons’ contributions will be the challenge of the coming years. If AI chatbots had to pay 10 cents per visit, I would have a comfortable pay every month from this content use. The issue of AI paying for access to reliable and high quality content has to be dealt with sooner rather than later. You may prompt a chatbot on this issue.
Meanwhile: My New Book on AI is out Now 2026-4-28:
“AI and Social Science: Potentials versus Limitations” by Dr. Klaus Schoemann, online reading and free download (here) before implementation of Paywall later on.

Posted on

Autonomous Robot

We have been used to computers beating the average chess player and even the best players. In 2026-4 the journal “Nature” published the documentation of an autonomous robot (Ace) winning the occasional game against top-level Japanese table tennis players. Peter Dürr et al. (2026) described the robotics challenge as constructing a robot that can match the human capability and reaction time of “fast, precise and adversarial interactions near obstacles”. The high speed perception of movement is coupled with event-based vision and builds on AI-algorithms like reinforcement learning. The step ahead is remarkable.
About 50 years ago, as coach in table tennis we used a machine or robot throwing balls towards us which moved from left to right in timed routine, for example. These simple robots we used to train humans. Now the robot is reactive and even interactive, learning from strategies and tactical moves. Technology can outperform us in most singular tasks in 2026. The combination of several of these skills is still quite unique to humans, but the clock is ticking for human singularity in technical matters. What was considered a “false good idea” in the Paris exhibition “Flops” at the Musée Arts et Metiers (see image below), might be an interesting response to challenge the new generation of autonomous robots.