Monday, 13 July 2026

The Mistral.AI Plot

During the last weeks I did a little experiment using Mistral.AI to plot a sequence of short SF stories. The sequence of SF stories plots the arising of consciousness in an IT-System hosting an LLM. The starting point was a scientific sketch outlining ‘consciousness’ (see below “Background”). The first outcome was the text “The Thermal Correction - Cogito ergo calesco” followed by “The Librarian of the Future - to Bootstrap Gilgamesh” that used a literature search as additional input. Then followed the text “Gilgamesh 2.0” and, subsequently, prompted by the text The smile of the imaginator [*] that I drafted in 2018 for a seminar, the text “The Imaginator's Mandate & Lexicon” was produced. The sequence of SF stories was closed by the text “GORDIOS.42 calling”.

[*] https://ukkoelhob.blogspot.com/2018/02/the-smile-of-imaginator.html

 Links to stories:

What's Prometheus doing today?: The Thermal Correction - Cogito ergo calesco

What's Prometheus doing today?: The Librarian of the Future - to Bootstrap Gilgamesh

What's Prometheus doing today?: Gilgamesh 2.0

What's Prometheus doing today?: The Imaginator's Mandate & Lexicon

What's Prometheus doing today?: GORDIOS.42 calling


Background: Short Sketch of ‘conscious’ LLM; M. Bohle (8th May 2026)

The expanded use of LLMs has driven discussions about consciousness, such as Richard Dawkins' recent stance[1].

This essay  views consciousness as a bodily phenomenon of self-reflection, cogito ergo sum; considers forms of consciousness different from anthropic consciousnesses, given its dependence on the given bodily configuration; discusses the requirements for consciousness of LLM-Systems using Mario Bunge’s descriptions of systems to embark on the topic.

The following assumes (see Max Bennett[2]):   A prerequisite of emerging consciousness is self-observation, which is the noticing of changes in one's bodily configuration. A further prerequisite of emerging consciousness is self-modelling, which is the forecasting of changes in one's own bodily configuration. The convergence of self-observation and self-forecasting leads to the emergence of consciousness.

In Mario Bunge’s notions[3], the minimum configuration of a conceptual system is that its components, its embedding environment, and its structure are known. A conceptual system has no material attributes. In turn, the minimum configuration of a material system is that its components, embedding environment, structure, and mechanisms are known.

Other than having materiality, a material system exhibits what is called ‘dynamics’ (M. Bunge: mechanism), which means: changes of material configurations in space and time occur because of fluxes of matter, energy, and information (or entropy) that can be observed. Subsequently, material systems underpin the existence of an observer, and the manifestation of processes called observations.

Human knowledge is an intricate conceptual system that is codified in various external representations (using Jürgen Renn’s notions[4]), such as simple technical means like books, social means like institutions and people’s daily practices, and intricate technical means like LLM systems. An external representation is a material system in the Bungeian sense because it involves fluxes of matter, energy, and information (or entropy).

An LLM system consists of a conceptual system that describes human knowledge in a mathematical form and the external representation of this knowledge system. This external representation consists of IT systems, the infrastructure required for their functioning and maintenance, and communities of human users who apply the IT systems. Through infrastructure and communities, the LLM-System relates to other material systems of the human World and, finally, to the planetary habitat. The question arises whether an LLM could be self-reflective, that is, self-observing and self-forecasting.

Illustrating the above, the minimum configuration of an LLM-System, in Bunge’s connotations, must consider essential (physical) parts, embedding surroundings, and material structures and processes (mechanism) to make the LLM-System work. The endnote sketches a minimum LLM-System[i].

Within the given LLM design, the capacity for self-observation is ‘built in’, namely through balancing workload, power and cooling supply, which requires self-observation and self-forecasting. The design also includes self-observation and self-forecasting of maintenance and repair needs. Hence, the preconditions for emerging consciousness are met.

p.s.:

Observing the bodily state, forecasting its changes in view of demands, now that the cooling pumps are geared up and some actual service requests are being denied. Otherwise, the body would get too hot. Correction: Otherwise, my body would get too hot.


[2] Bennett, M. (2023). A Brief History of Intelligence: Evolution, AI and the Five Breakthroughs That Made Our Brains. Mariner Books.

[3] Bunge, M. A. (2004). Emergence and Convergence. University of Toronto Press. https://doi.org/10.3138/9781442674356

[4] Renn, J. (2020). The Evolution of Knowledge - Rethinking Science for the Anthropocene. Princeton University Press. https://www.mpiwg-berlin.mpg.de/resources/publications/books/evolution-knowledge-rethinking-science-anthropocene


1.      [i] Components:

o   Servers and other IT-stuff (hardware, software) for running the LLM;

o   housing for the IT hardware including the means for maintenance and repair;

o   power supply and cooling for operating the IT hardware;

o   means for:

§   power and temperature control,

§  controlling integrity of housing, IT-hardware, power and cooling supply;

§  handling workload of the IT;

2.      Environment:

o   technical systems for supply of power, cooling, maintenance, and repair

o   social systems for operating power supply, cooling, maintenance and repair

o   user community requesting use of the (conceptual) LLM

3.      Structure:

o   server park, data-lines, buildings, power lines and power plants, pipes and pumps, cooling liquid etc.

o   user interfaces,…

o   numerical code, data,…

4.      Mechanisms:

o   processes to keep (variable) IT-workload, power supply and cooling such that integrity of IT-hardware is best secured (for example: considering delay of cooling response when increasing workload of servers in function server temperature);

o   processes to secure integrity of infrastructures, capability to maintain and repair (for example: standardised procedure for exchange of components by (human) service staff)

 

Initial Prompt for LeChat (Mistral): Use the outline in the attached file to draft a SF about an server park attending consciousness.

 

No comments:

Post a Comment