The Beginnings of a Hypothesis on How to Build Better Robots

• The industry of developing computer intelligence, machine learning and artificial intelligence is largely embedded in a particular version of cognitive science.
• This computing-industry often takes it ideas for memory, learning and development from the brand of evolutionary theory from a particular version of cognfitive science.
• This brand of evolutionary theory has at its heart an unresolved contradiction of how to account for development in evolution.
• Very briefly, this contradiction is the idea that an individual human and nonhuman life is subject to natural selection through its ecological context. But that humans are naturally economically incentivised in their actions and so exploit and subject their ecological context to meet their selfish needs.
• This is a theoretical contradiction. On one hand we are subjects of natural selection, on the other hand nature is our subject and we have this culture thing, but that also works like natural selection. Ingold unpacks this contradiction between Optimal Foraging Theory (OFT) and Economic Man (EM).
• A contradiction that fails to account for how a living system develops from childhood to adulthood.
• The way the associated ideology resolves this contradiction is by placing a categorical hard division between nonhumans and humans, the prior subject to OFT and the later to EM.
• In order to resolve the relation between the two along a linear evolutionary trajectory is to creates a third middle category of ‘primitive’ people who can sort of be nonhuman humans. We call them simple hunter-gatherers.
• This might all seem at this point quite far from computer algorithms. It’s not. The theory used to design and explain how information is transmitted, how learning occurs, how a system develops, what it means to be a user etc… are pretty much the same. Whether its an ecosystem ‘user’ or computer system user.

[Potential Insertion to Hypothesis, albeit something is still incorrect here I have yet to put my finger on]

• The dominant ideas we hold about hunter-gatherers emerge as contention between Hobbes and Rousseau as either barbaric primitives or noble savages. These two different origin stories underpinning different dimensions of conservative and liberal political histories and values.
• These two traditions arguably being an iteration of EM and OFT. Economic man the savage dominator and exploiter, who then needs strong control to keep order. Hence, Hobbes Sovereign King. Or Rousseau’s man at harmony with nature, an optimal forager, until the rise of the mutant tyrannt who exploits everyone. Hence, Rousseau’s democratc proposals to address tyrants.
• In both cases it is exploitation that is designated as what divides modern man from nature and primitive man. It is posited as what it truly means to be human.
• This is precisely the commonality between OFT and EM. They both conceptualise humans and nonhumans as relating to their ecology as exploitation. As one thing taking something from another thing. As transactional transmission.

[Insertion Ends/]

• It is the “naturalist perspective that emerges with a view of all relations, material or otherwise as transactional. It is part of the idea of a scientific experimental process, a relationship, as self-vindicating. That is that one receives knowledge/resource, then applies it toward achieving an aim, and if the knowledge/resource is correct, it increases your ‘fitness’ and the aim is achieved. This is a romanticized view of the experimental process, of human-environmental relations and of knowledge as transactional transmission” (Me in 2018: 82, fn 1)
• What does this have to do with computing today and your life? A lot.
• Basically algorithms, what is today called AI and associated infrastructures are increasingly designed according to the principle of knowledge/materials as informational resources to be transmitted around.
• This idea is rooted in an OFT/EM understanding of life which also has to create a primitive or ‘other’ to resolve the inherent contradiction at its heart. In addition the contradiction is one that does not allow for an accurate theory of how development and learning take place.
• If it did we would not consider AI today as essentially a hyper exaggerated category enforcer, expecting life to mould to it rather than with it.
• If we resolved this contradiction a very different AI could emerge.
• One in some sense more akin to life rather than firstly an exercise in false portrayal (Turing Machines) followed by calculators that spy and mine you (surveillance capitalism) and Stupid (Big) Data. All designed in a way that bastardises the potential of AI.
• In short computing has come to reflect a poor science of what it means to be human combined with donkey loads of cash.