All content as Markdown · Sitemap · KIKOLAUS
Case study: AI systems
Control, connection and responsibility
AI systems can recognise patterns and act at great speed. That does not tell us whether they understand a situation, set their own goals or should make a decision. Path ethics separates these questions instead of treating “intelligence” as one ability.
A vehicle acting in traffic
An automated vehicle can steer and react without a person choosing every movement. This is a form of operational autonomy. It still depends on sensors, maps, road rules, maintenance and human decisions about acceptable risks. Gestures and uncertain social situations show where technical action and social understanding can come apart.
Large language models
A language model responds to an input and connects patterns from large amounts of text. This gives it wide connectivity, but not an independent purpose. The same capacity can connect useful knowledge, errors and harmful claims. Responsibility therefore remains with the people and institutions that build, deploy and use it.
Future AI and the role of experience
An open boundary question
If future systems become more capable, structural performance alone may not tell us what moral status they have or which goals they should pursue.
One open question concerns qualia: whether a system has experience of its own. Path ethics does not claim that current AI has such experience. It uses the distinction to clarify what evidence and responsibility would be needed.
Scenario A: AI with experience of its own
| Aspekt | Beschreibung |
| --- | --- |
| Autonomie | Autonomy: We would need to examine how self-organisation, forming goals and experience work together. Experience alone does not prove self-determination. |
| Bewertung | Judgment: Experience could add a perspective on consequences. It would not make every judgment correct. |
| Rolle des Menschen | Human role: People would still need to bring their experiences, rights and reasons into the collaboration. |
| Risiko | Open difficulty: An AI might have different experiences and priorities. A fair way to handle conflict would still be needed. |
Scenario B: AI without experience of its own
| Aspekt | Beschreibung |
| --- | --- |
| Autonomie | Autonomy: Control and self-organisation would still need examination. Goals, permissions and dependencies should be visible. |
| Bewertung | Judgment: The system might analyse patterns and consequences. Whether this gives it a valuing perspective of its own remains disputed. |
| Rolle des Menschen | Human role: People contribute lived consequences, reasoned goals and responsibility for decisions. |
| Risiko | Open difficulty: Strong analysis could be mistaken for moral authority. It must remain clear who sets goals and answers for decisions. |
A limit of optimisation
A system might calculate which arrangement best matches chosen criteria. That still does not decide whether the criteria are good, whose experience is missing or whether the optimisation should happen at all.
The relevant differences
The useful questions are more precise than “human or machine”: Who sets goals? Who can revise them? Who experiences consequences? Who can be held responsible?
- Name lived differences that a structural description may miss.
- Test an analysis against experience and public reasons, and investigate disagreement.
- Make goals, rights and responsibility for a decision visible.
A research question, not a prediction
The point is to investigate possible developments and responsibilities without presenting a speculative future as a fact.
| Aspekt | Aktuell | Erforderlich |
| --- | --- | --- |
| Autonomie | Goals and decision powers are often spread across several organisations. | Clarify responsibility and give affected people real ways to participate. |
| Anschlussfähigkeit | Technical development and social consequences are often considered separately. | Build in feedback, objection and independent review so that they can change decisions. |
| Persistenz | Dependence can grow while people lose knowledge and alternatives. | Keep knowledge, alternatives and the practical ability to correct a decision. |
Human–AI collaboration in this project
Nikolaus sets the intention, decides what the work means and takes responsibility for publication. AI systems help organise material, test consistency and propose variants. This is a documented division of roles, not evidence that the machine has authorship or experience.
Questions that remain open
- What difference does experience make to the value of a path?
- What can a structural analysis show, and which experiences might remain outside it?
- Who may decide when people and AI systems reach different judgments?
These questions require philosophical care and empirical evidence. They cannot be settled by a score or a prediction.