Dear SoTA,

Optimism is found in strange places during the early stages of 2026: amidst airstrikes, commentators cheer for change; among political gamesmanship, voters share hopes for progress; as human workforces are laid off en masse, technologists laud the potential of AI to solve our collective problems.

Its in this final point that we see the widest gap between the current reality and the dearly hoped for outcome that dominates popular discourse. It is assumed a priori that AI technology is good and helpful, and that more integrated systems are more good and more helpful. But,

"to whom does this benefit accrue?"
is the question we must ask aloud, and

"what is the result for humanity if we continue in this way?"
we may ask ourselves, under our breath, in quiet rooms with few around.

We propose an inquiry first and a solution second to this matter; an economic system that drives value creation in the AI age back to the humans that create it. We highlight the urgent need for more equitable distribution of profit from and access to frontier AI systems, the demand for human representation in an LLM-processed reality, and suggest how to expand the potential for self-actualisation despite collapsing possibility space.

As we push closer to fully automated societies, it is essential that we consider humanity and our role within it with sober realism. AI systems are trained by us, the collective humanity, willingly or otherwise. Thus, one could expect a symbiotic relationship, one of repaying that which was given, which nature has proven time and time again is the only truly scalable system. Instead, we are seeing an extractive relationship. We are seeing wholesale replacement and disenfranchisement of the human worker by the very systems they feed. This parasitic interaction between machine and the human mind inevitably leads to a dead host and a mad machine.

The extraction of knowledge from the human mind is nothing new. Early innovators of the Industrial Revolution carefully observed the expert knowledge of the guilds in England: the precise action of the craftsman's knife, the cadence of the weaver's movement, the measured power of a mason's hammer swing. They embedded these qualities into the earliest industrial machines. From Cartwright's Mechanical Loom to Babbage's Counting Machine, the knowledge that informed the machine was captured from human minds.

To capture knowledge is to abstract it, to condense or process the ephemeral "knowing of something" into a discrete object. From the mind, to the mouth, to paper, to image and now to code, humans have steadily increased the scale of knowledge representation. What fidelity we may have lost along the way can be debated but one thing is clear: the amount of knowledge available to human and machine alike is growing exponentially. We hope to inspire a system that supports both.

For all our grand inventions, many argue that mankind's greatest innovation is the market, the hidden hand that shapes worlds and determines destinies. While the author disagrees with the existing market function as an ideal value exchange mechanism, it is nonetheless the board on which we are all playing. As such, we envisage a rewriting of the rules, the introduction of a new playing piece: knowledge as an asset.

If we can codify knowledge into an asset, despite its abstractions, it can unlock a new class of labour: the knowledge worker. The knowledge worker has always been present in society and is yet to be fully represented in economic practise. Karl Marx spoke of the General Intellect in Grundrisse, the collective knowledge of society as a force of production. He presciently suggests that the development of fixed capital (any kind of physical asset that is used repeatedly in production) indicates to what degree the general intellect acts as a direct force of production. In the case of AI, the massive scale of compute farms is directly proportional to the amount of collective knowledge employed and utilised.

As of yet however, there are few manifestations of knowledge that generates value for the thinker. We mostly use IP as a method to capture the value of the intellectual input, but this is deeply flawed and must be improved. Knowledge wants to be known and IP directly counters this natural flow of information.

We also see an opportunity for the general population in the AI systems that threaten to doom it. We see knowledge, the lifeblood of artificial intelligence and the product of the human mind, as an avenue for fairly distributing the gains of the productivity boom we are about to experience.

If the people decide to adopt sovereign systems, digital infrastructure capable of allowing self-custody, then the knowledge which is currently being extracted may be withheld. The value that is created can belong to the person whose mind dreamt or processed it into digital being. In a world of abundance, there is no difference between learning, work and play - it is all research. And so, if we can correctly structure knowledge, distribute its ownership and incentivise its creation, humans may very well thrive in a post-scarcity world. Instead of the encroaching role as host to parasitic machine, to thrive as humans, assisted by their creations.

Watch a presentation at Imperial College London on this matter, from 2024:

I leave this matter for a future letter but if you are interested in our work, please email me joshua@desci.world or visit us at bonfires.ai to begin building your own knowledge asset.

Kind Regards,

Joshua

Dear SoTA,

Optimism is found in strange places during the early stages of 2026: amidst airstrikes, commentators cheer for change; among political gamesmanship, voters share hopes for progress; as human workforces are laid off en masse, technologists laud the potential of AI to solve our collective problems.

Its in this final point that we see the widest gap between the current reality and the dearly hoped for outcome that dominates popular discourse. It is assumed a priori that AI technology is good and helpful, and that more integrated systems are more good and more helpful. But,

"to whom does this benefit accrue?"
is the question we must ask aloud, and

"what is the result for humanity if we continue in this way?"
we may ask ourselves, under our breath, in quiet rooms with few around.

We propose an inquiry first and a solution second to this matter; an economic system that drives value creation in the AI age back to the humans that create it. We highlight the urgent need for more equitable distribution of profit from and access to frontier AI systems, the demand for human representation in an LLM-processed reality, and suggest how to expand the potential for self-actualisation despite collapsing possibility space.

As we push closer to fully automated societies, it is essential that we consider humanity and our role within it with sober realism. AI systems are trained by us, the collective humanity, willingly or otherwise. Thus, one could expect a symbiotic relationship, one of repaying that which was given, which nature has proven time and time again is the only truly scalable system. Instead, we are seeing an extractive relationship. We are seeing wholesale replacement and disenfranchisement of the human worker by the very systems they feed. This parasitic interaction between machine and the human mind inevitably leads to a dead host and a mad machine.

The extraction of knowledge from the human mind is nothing new. Early innovators of the Industrial Revolution carefully observed the expert knowledge of the guilds in England: the precise action of the craftsman's knife, the cadence of the weaver's movement, the measured power of a mason's hammer swing. They embedded these qualities into the earliest industrial machines. From Cartwright's Mechanical Loom to Babbage's Counting Machine, the knowledge that informed the machine was captured from human minds.

To capture knowledge is to abstract it, to condense or process the ephemeral "knowing of something" into a discrete object. From the mind, to the mouth, to paper, to image and now to code, humans have steadily increased the scale of knowledge representation. What fidelity we may have lost along the way can be debated but one thing is clear: the amount of knowledge available to human and machine alike is growing exponentially. We hope to inspire a system that supports both.

For all our grand inventions, many argue that mankind's greatest innovation is the market, the hidden hand that shapes worlds and determines destinies. While the author disagrees with the existing market function as an ideal value exchange mechanism, it is nonetheless the board on which we are all playing. As such, we envisage a rewriting of the rules, the introduction of a new playing piece: knowledge as an asset.

If we can codify knowledge into an asset, despite its abstractions, it can unlock a new class of labour: the knowledge worker. The knowledge worker has always been present in society and is yet to be fully represented in economic practise. Karl Marx spoke of the General Intellect in Grundrisse, the collective knowledge of society as a force of production. He presciently suggests that the development of fixed capital (any kind of physical asset that is used repeatedly in production) indicates to what degree the general intellect acts as a direct force of production. In the case of AI, the massive scale of compute farms is directly proportional to the amount of collective knowledge employed and utilised.

As of yet however, there are few manifestations of knowledge that generates value for the thinker. We mostly use IP as a method to capture the value of the intellectual input, but this is deeply flawed and must be improved. Knowledge wants to be known and IP directly counters this natural flow of information.

We also see an opportunity for the general population in the AI systems that threaten to doom it. We see knowledge, the lifeblood of artificial intelligence and the product of the human mind, as an avenue for fairly distributing the gains of the productivity boom we are about to experience.

If the people decide to adopt sovereign systems, digital infrastructure capable of allowing self-custody, then the knowledge which is currently being extracted may be withheld. The value that is created can belong to the person whose mind dreamt or processed it into digital being. In a world of abundance, there is no difference between learning, work and play - it is all research. And so, if we can correctly structure knowledge, distribute its ownership and incentivise its creation, humans may very well thrive in a post-scarcity world. Instead of the encroaching role as host to parasitic machine, to thrive as humans, assisted by their creations.

Watch a presentation at Imperial College London on this matter, from 2024:

I leave this matter for a future letter but if you are interested in our work, please email me joshua@desci.world or visit us at bonfires.ai to begin building your own knowledge asset.

Kind Regards,

Joshua