Monday, December 18, 2023
008 - Lessons from Artificial Intelligence (Part VI)
In the sixth and final episode of this subseries, I go into further detail about the similarities between artificial neural networks and of the human mind, and how the same dynamics of learning are used to influence our behavior and perception of reality.
Duration: 00:07:13
Episode Transcript
Intro
Good morning. You’re listening to the Reflections in Beige Podcast, hosted by Michael LeSane. The date is Monday, December 18th, 2023.
I must apologize for the long delay in releasing a new episode. I was exceedingly busy with my job, and with a series of very important deadlines on the horizon, a myriad of other matters in my life were neglected wholesale. But for bureaucratic reasons independent of those deadlines, I parted with the company in the end, and am once again forging my own professional path.
So for now, I’m back. Let’s continue.
Prelude to Void
Reflecting on my experiences working with artificial neural networks, I’ve always said that it ruined humanity for me.
It distills the process of learning, of conditioning, of programming to its core essence: observation through the senses, inference and decision – both conscious and latent – based on those observations, and reorientation of perspective or mode of operation based on feedback.
Consider the use of flash cards when studying foreign languages, for instance. You pull a card from the deck, read a word or phrase, anticipate its translation, flip the card over, and read the translation. Your understanding of this word or phrase is either confirmed, in which case the card is removed from the deck for the study session, or the card is shuffled back into the deck for you to review again.
The flash card study session may be considered “complete” when you’ve demonstrated comprehension of all the cards, front and back. In other words, you have reconfigured your mind to retain all the linguistic associations conveyed by these cards: word to word, phrase to phrase, native tongue to foreign tongue, and foreign tongue to native tongue.
If this learning process was expressed in terms of the flow of a neural network, the exposure to one side of the card would be the input stimuli. The attempt to recall the word or phrase one side is associated with on the other side would be the forward pass. The other side of the card is the feedback, the margin of error is how different it is from the initial prediction, and backpropagation is the passive process of the mind iteratively adjusting its understanding thereafter to more efficiently and accurately recall the association the next time it is exposed to the pattern in question.
The frustration or other emotional disturbances associated with difficult lessons, failures of undertakings, or more generally outcomes which diverge from expectations are effectively manifestations of learning (see: backpropagation) with consciousness of loss with respect to some emotionally vested interest. Loss of time, energy, other resources, or loss of tangible or intangible ground to an undesirable ends.
The Real
Each and every day in life, we are exposed to an endless stream of patterns to make sense of or to inform how we should proceed, and feedback which either vindicates our understanding and decisions or which might behoove us to reconsider what we think or do in response to things. Some stimuli and feedback, most even, might even be filtered out altogether as noise, as the alternative to this is effectively being overwhelmed by sensory overload. Neurochemicals like serotonin serve to regulate this as well, which is why its deficiency tends to correlate with higher levels of anxiety.
This all serves to form our understanding of the world and our relationship with it.
The Virtual
These patterns and feedback come not just in the form of real, material, firsthand experiences, but also in more virtual forms.
One example of this is information relayed through interpersonal communication: educational instruction, casual conversation, small talk, gossip, et cetera. Another example of this is through interactive forms of entertainment like games – video games, board games, sports, and other forms of play. Finally, you have more passive forms of consumed information, like literature, press, television, and film.
All of these forms of virtualized experiences frame realities of different scopes and governed by different laws, some fixed like the rules of board games; some fluid like absurdist literature, comedy television, or supernatural horror films. Most are somewhere in between.
When knowingly immersing ourselves in fictional virtual realities and recreational games, we generally suspend belief, approaching it like a microcosm with elements of reality but with intrinsically distinct characteristics.
The Simulation
However, we also unwittingly immerse ourselves in fictional realities and recreational games, without suspending belief. Simulations, if you will. We approach what is at best an incomplete microcosm and at worst a flawed imitation of reality governed by different laws, presented as though it is reality. Our understanding of the world is contextualized – or trained, if you will – by distorted stimuli and feedback to the point of overfitting, and we are deliberately instilled with an emotional vested interest in certain outcomes, to spur our involvement towards certain ends or alternately to provoke a certain range of responses when undesired outcomes unfold.
Examples of this include the lies and deception of trusted acquaintances; the misrepresentation of subject matter in nonfiction literature, publications, films based on true stories, and “informative” television programming like the news; information sometimes provided in good faith but stemming from corrupted sources of truth; and information coming from sound or rigorous sources but framed in a dishonest manner.
Sometimes trusted sources of information are corrupted, but so too are sources of information deemed corrupted sometimes more truthful. Sometimes this is inadvertent; sometimes it is deliberate.
The motives for these manipulations of our perceptions of reality may vary – it could be social, commercial, financial, or political in nature. It could even be scientific, as exemplified by the large scale experimentation on user sentiment by data scientists at Facebook in 2012.
Conclusion
At the end of the day, our worldview is a mesh of associations constructed from many different sources and experiences, and we are largely united by shared truths and shared delusions.
While it is impossible to be absolutely correct in our assessment or understanding of the stimuli we encounter or more generally the realities we navigate, it behooves us to remain cognizant of the basic nature of how our minds work and how stimuli works on our minds.
In doing so, we can make more informed decisions about what information we consume and how we consume that information, which ultimately dictates our decisions, the decisions of others, our relationships with each other, and our outcomes in life.
Let that be the lesson for today. This concludes my series recounting experiments with artificial neural networks and what I took away from it all. I hope that you’ve found it to be interesting food for thought, and that it might spark new curiosities to explore and indulge in.
Thank you for listening and sharing, and have a good day.
Good morning. You’re listening to the Reflections in Beige Podcast, hosted by Michael LeSane. The date is Monday, December 18th, 2023.
I must apologize for the long delay in releasing a new episode. I was exceedingly busy with my job, and with a series of very important deadlines on the horizon, a myriad of other matters in my life were neglected wholesale. But for bureaucratic reasons independent of those deadlines, I parted with the company in the end, and am once again forging my own professional path.
So for now, I’m back. Let’s continue.
Prelude to Void
Reflecting on my experiences working with artificial neural networks, I’ve always said that it ruined humanity for me.
It distills the process of learning, of conditioning, of programming to its core essence: observation through the senses, inference and decision – both conscious and latent – based on those observations, and reorientation of perspective or mode of operation based on feedback.
Consider the use of flash cards when studying foreign languages, for instance. You pull a card from the deck, read a word or phrase, anticipate its translation, flip the card over, and read the translation. Your understanding of this word or phrase is either confirmed, in which case the card is removed from the deck for the study session, or the card is shuffled back into the deck for you to review again.
The flash card study session may be considered “complete” when you’ve demonstrated comprehension of all the cards, front and back. In other words, you have reconfigured your mind to retain all the linguistic associations conveyed by these cards: word to word, phrase to phrase, native tongue to foreign tongue, and foreign tongue to native tongue.
If this learning process was expressed in terms of the flow of a neural network, the exposure to one side of the card would be the input stimuli. The attempt to recall the word or phrase one side is associated with on the other side would be the forward pass. The other side of the card is the feedback, the margin of error is how different it is from the initial prediction, and backpropagation is the passive process of the mind iteratively adjusting its understanding thereafter to more efficiently and accurately recall the association the next time it is exposed to the pattern in question.
The frustration or other emotional disturbances associated with difficult lessons, failures of undertakings, or more generally outcomes which diverge from expectations are effectively manifestations of learning (see: backpropagation) with consciousness of loss with respect to some emotionally vested interest. Loss of time, energy, other resources, or loss of tangible or intangible ground to an undesirable ends.
The Real
Each and every day in life, we are exposed to an endless stream of patterns to make sense of or to inform how we should proceed, and feedback which either vindicates our understanding and decisions or which might behoove us to reconsider what we think or do in response to things. Some stimuli and feedback, most even, might even be filtered out altogether as noise, as the alternative to this is effectively being overwhelmed by sensory overload. Neurochemicals like serotonin serve to regulate this as well, which is why its deficiency tends to correlate with higher levels of anxiety.
This all serves to form our understanding of the world and our relationship with it.
The Virtual
These patterns and feedback come not just in the form of real, material, firsthand experiences, but also in more virtual forms.
One example of this is information relayed through interpersonal communication: educational instruction, casual conversation, small talk, gossip, et cetera. Another example of this is through interactive forms of entertainment like games – video games, board games, sports, and other forms of play. Finally, you have more passive forms of consumed information, like literature, press, television, and film.
All of these forms of virtualized experiences frame realities of different scopes and governed by different laws, some fixed like the rules of board games; some fluid like absurdist literature, comedy television, or supernatural horror films. Most are somewhere in between.
When knowingly immersing ourselves in fictional virtual realities and recreational games, we generally suspend belief, approaching it like a microcosm with elements of reality but with intrinsically distinct characteristics.
The Simulation
However, we also unwittingly immerse ourselves in fictional realities and recreational games, without suspending belief. Simulations, if you will. We approach what is at best an incomplete microcosm and at worst a flawed imitation of reality governed by different laws, presented as though it is reality. Our understanding of the world is contextualized – or trained, if you will – by distorted stimuli and feedback to the point of overfitting, and we are deliberately instilled with an emotional vested interest in certain outcomes, to spur our involvement towards certain ends or alternately to provoke a certain range of responses when undesired outcomes unfold.
Examples of this include the lies and deception of trusted acquaintances; the misrepresentation of subject matter in nonfiction literature, publications, films based on true stories, and “informative” television programming like the news; information sometimes provided in good faith but stemming from corrupted sources of truth; and information coming from sound or rigorous sources but framed in a dishonest manner.
Sometimes trusted sources of information are corrupted, but so too are sources of information deemed corrupted sometimes more truthful. Sometimes this is inadvertent; sometimes it is deliberate.
The motives for these manipulations of our perceptions of reality may vary – it could be social, commercial, financial, or political in nature. It could even be scientific, as exemplified by the large scale experimentation on user sentiment by data scientists at Facebook in 2012.
Conclusion
At the end of the day, our worldview is a mesh of associations constructed from many different sources and experiences, and we are largely united by shared truths and shared delusions.
While it is impossible to be absolutely correct in our assessment or understanding of the stimuli we encounter or more generally the realities we navigate, it behooves us to remain cognizant of the basic nature of how our minds work and how stimuli works on our minds.
In doing so, we can make more informed decisions about what information we consume and how we consume that information, which ultimately dictates our decisions, the decisions of others, our relationships with each other, and our outcomes in life.
Let that be the lesson for today. This concludes my series recounting experiments with artificial neural networks and what I took away from it all. I hope that you’ve found it to be interesting food for thought, and that it might spark new curiosities to explore and indulge in.
Thank you for listening and sharing, and have a good day.