How Neuroscience and AI Play Politics with our Brains

Author(s): Dr. Adam Hart

Originally published on Towards AI.

“Why should you buy the computational theory of the mind? Because it has solved millennia-old problems in philosophy, kicked off the computer revolution, posed the significant questions of neuroscience, and provided psychology with a magnificently fruitful research agenda.”

Steven Pinker, How the Mind Works, p77, 1997

“It is the political retreat from Enlightenment rationalism, from a belief in human agency, and from ideas of moral progress, that has opened up the space for a mechanistic view of Man”

Kenan Malik, “Man, Beast and Zombie — What science can and cannot tell us about human nature” p390, 2000.

In 1997 self-confirmed atheist, advocate for the computational theory of the mind (CTM), and Harvard Professor Dr. Steven Pinker perhaps could not dream that in ~30 years time the achievements of DeepMind and AlphaGo Zero would even occur.

Yet they have, and due to the confluence of neuroscience and a branch of computer science, AI, that has embraced the computational theory of the mind wholeheartedly.

Neuroscience and the different types of deep neural networks that are being pursued as AI, of which reinforcement learning is the one most spectacularly being used to effect, work very well together as the talented CEO of DeepMind Dr. Demis Hassabis’s knows since his work crosses both fields.

Their recent paper in Nature regards the role of dopamine with reinforcement learning.

DeepMind, an AI R&D division of Google acquired for £400M in 2014, is using studies of mice to reverse engineer the mice’s brain’s functions into code that can learn.

Why are they interested in this?

Part of their agenda must be to strive towards the monetization of an artificial human general intelligence to replace humans with code that learns. To do this, the 400 odd people there study how animals and humans think and learn and strive to factor that into code.

Neuroscience asserts we are motivated to learn through dopamine. Specifically, the dopamine reward prediction error model is said to govern reinforcement learning, which in turn may be said: “to underlie much of human and animal behavior.”

Dopamine is a neurotransmitter that “plays a role in how we feel pleasure. It’s a big part of our uniquely human ability to think and plan. It helps us strive, focus, and find things interesting.” [1]

Dopamine, DeepMind asserts, at least in mice, provides a series of rewards and punishments whereby the brain supposedly calculates the maximum pleasurable or the least painful ‘hit’ from a suite of possible future outcomes.

Apart from DeepMind, significant use of ML and AI [2] is occurring across the globe, examples include:

1. Preferences specification — Google and Facebook advertising topics are automatically populated based on browsing/feed meta-data using ML;
2. Discovery of Identity — ClearView AI used by US police departments to identify suspects or victims using visual pattern recognition;
3. Surveillance — using facial pattern recognition or even weaponized drones;
4. Optimization of inputs/outputs — DeepMind sold parent Alphabet code to optimize Google data center air conditioning reducing power consumption by 40%; ML photography optimization in iPhone 12 and Pixel 4;
5. Novelty — Mitsuku chatbot; enlighten illumination engine for video games;
6. Fakery — MelNet and DeepFakes;
7. Hacking — malicious tweeting to mislead or even hacking AI code itself;
8. Augmentation of human capability — Boston Dynamics Spot, IBM Watson; and
9. *Replacement of human capability — DeepMind’s AlphaGo Zero; Soul Machines; RealDollX; UneeQ digital humans; Tesla Autopilot; Tesla Smart Summon; Squirrel AI automated teaching.

This list shows the breadth and depth of AI and is just an entree. Every significant Tech company and University is investing in this domain, many with the firm belief that the computational theory of the mind is true, proper and correct, because of the results they have been getting.

Questioning the Computational Theory of the Mind.

As DeepMind’s research into the ‘canonical’ dopamine reward prediction error model indicates, they wish to or already have factored this belief into reinforcement neural network code. Classical CTM holds that a mind is just a Turing Machine. Also, while aligned physicalist philosophers such as the American Dr. Dan Dennett assert that the true nature of consciousness is fragmentary and consciousness is the computation, and the philosophy of the mind that many software researchers use is founded in the analytical (Anglo-American) tradition, there is an opposing Continental view.

That view allows me to question, is it right that Karl Popper’s lauded scientific method has leaked across from the practice of physics and chemistry to the study of the human mind? If I accept that it isn’t correct to do so because the human mind and our societies are not a lump of coal, am I, therefore, permitted to not believe in the physicalist West Coast analytic philosophy?

If so, then this opens up the opportunity to view the computational theory of the mind differently, to view it as an emergent property of a discourse that wishes to pin down humanity at all costs at it’s most fundamental level for purposes of control and prediction, which after all is the core purpose of chemistry and physics too. Excepting that physics and chemistry want to do this for inert objects, and neuroscience and AI researchers want to do this to humans, to turn general humanity into predictable objects.

Secondly, that the joining of a computational theory of the mind with ‘proven’ medical facts that the mind learns through a dopamine optimization strategy of rewards for future actions is intended to assert a superordinate yet mechanistic relation between their theory of control and prediction and human potentiality, human agency.

Thirdly, I can question this because the computational theory of the mind and neuroscience is purposely researching and inventing AI tech to replace humans.

So the agenda is clear. Clever theories (not truths) advocated by tenured Professors purporting to explain the human brain are being used to create and deploy tech that seeks to replace and disenfranchise populations of humans.

CTM isn’t a theory of the mind. Instead, it is a theory of human political control coupled with autonomous tools that measure images, detect patterns of behavior, and assert answers to questions as truths from code with black box (DNN) logic that cannot be explained by people.

Through supposedly superior scientism, communities of scientists aided in some instances by surveillance-oriented governments like China (cf. Social Credit Scoring) or Australia (cf. The Capability) are hell-bent on proving the correctness of their belief system by imposing it on unwitting societies.

Innocent populations intertwined into a global AI experiment using social media, ubiquitous smart devices, and cameras as the channels.

What else this is, is a longitudinal shift of belief against the divine, and a push towards atheism and science as the superior global discourse that accompanies high and ultra-high net wealth economics.

In 2013, Professor Pinker was awarded a Richard Dawkins Award from the Atheist Alliance of America.

How is the dopamine reward prediction error model relevant to the inspirational works of Michelangelo, Mozart, or Handel?

It’s not. But it is relevant to predicting and controlling behavior through a series of rewards and penalties, it is relevant to computational strategies for code to learn Go, and it will become increasingly relevant to the government of peoples as they too adopt increasingly pervasive automated reinforcement learning strategies, such as the Australian governments monitoring and restriction of social security payment spending to authorized products and services, which exclude alcohol, gambling or cash.

Whether you believe in the divine or not, what matters in a free society that a multiplicity of thought systems, beliefs, and discourses are permitted. Where one group is aiming to prove the other wrong that is firmly a relation of power and control. The Dali Lama doesn’t seek to force you to believe that Buddhism is the true way, neither should the AI research community seek to inject monetized code across the globe in many contexts of everyday life that is predicated on an politico-mechanistic lens on humanity simply for its efficacy at tracking, tricking, controlling and even replacing humans.

It seems the brain has become the new political bargaining tool.

Footnotes

[1] https://www.webmd.com/mental-health/what-is-dopamine#1

[2] There’s a lot of confusion in the media intermixing machine learning (ML), which is fundamentally an accepted statistical tool, with the nascent artificial intelligence (AI) advances. ML uses probability distributions and associated measurements and techniques, while AI is based on neural networks using theories of reinforcement learning or other models that reference back to the computational theory of the mind.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI