Painful intelligence:
What AI can tell us about
human suffering
Aapo Hyvärinen
University of Helsinki
Version 1.01
23rd June 2022
Book home page
This book uses the modern theory of artificial intelligence (AI) to understand human suffering or mental pain. Both
humans and sophisticated AI agents process information about the world in order to achieve goals and obtain
rewards, which is why AI can be used as a model of the human brain and mind. This book intends to
make the theory accessible to a relatively general audience, requiring only some relevant scientific
background.
The book starts with the assumption that suffering is mainly caused by frustration. Frustration means the
failure of an agent (whether AI or human) to achieve a goal or a reward it wanted or expected.
Frustration is inevitable because of the overwhelming complexity of the world, limited computational
resources, and scarcity of good data. In particular, such limitations imply that an agent acting in the
real world must cope with uncontrollability, unpredictability, and uncertainty, which all lead to
frustration.
Fundamental in such modelling is the idea of learning, or adaptation to the environment. While AI uses
machine learning, humans and animals adapt by a combination of evolutionary mechanisms and ordinary
learning. Even frustration is fundamentally an error signal that the system uses for learning. This book
explores various aspects and limitations of learning algorithms and their implications regarding
suffering.
At the end of the book, the computational theory is used to derive various interventions or training methods
that will reduce suffering in humans. The amount of frustration is expressed by a simple equation which
indicates how it can be reduced. The ensuing interventions are very similar to those proposed by Buddhist and
Stoic philosophy, and include mindfulness meditation. Therefore, this book can be interpreted as an
exposition of a computational theory justifying why such philosophies and meditation reduce human
suffering.
Preface
1 Introduction
Investigating intelligence by constructing it
Is the brain a big computer?
Machine learning as analogue to evolution
Can an AI actually suffer?
Intelligence is painful—overview of this book
I Suffering as error signalling
2 Defining suffering
Medical definitions of pain
Medical and psychological definitions suffering
Ancient philosophical approaches to suffering
Two main kinds of suffering
Using the pain system for broadcasting errors
3 Frustration due to failed plan
Agents, states, and goals
Planning action sequences, and its great difficulty
Frustration as not reaching planned goal
Defining desire as a goal-suggesting mechanism
Intention as commitment to a goal
Heuristics can help in planning
4 Machine learning as minimization of errors
Neurons and neural networks
Finding the right function by learning
Learning as minimization of errors
Gradient optimization vs. evolution
Learning associations by Hebbian rule
Logic and symbols as an alternative approach
Emergence of unexpected behaviour
5 Frustration due to reward prediction error
Maximizing rewards instead of reaching goals
Learning to plan using state-values and action-values
Frustration as reward loss and prediction error
Expectations or predictions are crucial for frustration
Unexpected implications of state-value computation
Evolutionary rewards as obsessions
Reward maximization is insatiable
6 Suffering due to self-needs
Self as long-term performance evaluation
Self as self-preservation and survival
Self as desires based on internal rewards
Uncertainty, unpredictability, and uncontrollability as internal frustration
Fear, threat, and frustration
7 Fast and slow intelligence and their problems
Fast and automated vs. slow and deliberative
Neural network learning is slow, data-hungry, and inflexible
Using planning and habits together
Advantages of categories and symbols
Categorization is fuzzy, uncertain, and arbitrary
The many faces of frustration: Summarizing the mechanisms of suffering
II Origins of suffering: uncontrollability and uncertainty
8 Emotions and desires as interrupts
Computation is one aspect of emotions
Emotions interrupt ongoing processing
Desire as an emotion and interrupt
Emtions include hard-wired action sequences
How interrupts increase suffering
Emotions are boundedly rational
9 Thoughts wandering by default
Wandering thoughts and the default-mode network
Wandering thoughts as replay and planning
Experience replay focuses on reinforcing events
Replay exists in rats, humans, and machines
Wandering thoughts multiply suffering
10 Perception as construction of the world
Vision only seems to be effortless and certain
Perception as unconscious inference
Prior information can be learned
Illusions as inference that goes wrong
Attention as input selection
Subjectivity and context-dependence of perception
Reward loss as mere percept
Ancient philosophers on perception
11 Distributed processing and no-self philosophy
Are you really in control?
Necessity of parallel and distributed processing
Central executive and society of mind
Control as mere percept of functionality
Philosophy of no-self and no-doer
12 Consciousness as the ultimate illusion
Information processing vs. subjective experience
The computational function of human consciousness
The origin of conscious experience
Why is simulated suffering conscious?
Self vs. consciousness
Nothing is real?
III Liberation from suffering
13 Overview of the causes and mechanisms
Why there is (so much) suffering
Cognitive dynamics leading to suffering
An equation to compute frustration
14 Reprogramming the brain to reduce suffering
Reducing expectation of rewards
Reducing certainty attributed to perception and concepts
Reducing self-needs
Reducing desire and aversion
How far should reducing desires and expectations go?
15 Retraining neural networks by meditation
Contemplation as active replay
Mindfulness meditation as training from a new data set
Speeding up the training
Reducing interrupting desires
Emptying the mind and reducing simulation
Attitude of acceptance
Metacognition and observing the nature of mind
Letting go and relaxation as unifying principles
16 Epilogue
Bibliography
Copyright ©2022 Aapo Hyvärinen. All rights reserved.
Distribution allowed as per Creative Commons Attribution-Noncommercial-NoDerivatives (CC BY-NC-ND) License.
[next]