Are We Living With AI? Exploring the Evolution of Neural Networks
Written on
Chapter 1: The Concept of AI in Everyday Life
What if the actions of humanity over countless generations have collectively shaped a global Neural Network? This thought challenges our conventional understanding of Artificial Intelligence, which we typically view as a construct of human design.
Could it be that we have inadvertently been operating a vast Neural Network throughout history? This notion raises significant questions about how such a system might have influenced our existence. Neural Networks—whether in the form of computer simulations or the biological neurons in our own brains—represent mechanisms for processing information in specific patterns. Remarkably, it appears that our daily activities have indeed contributed to the functioning of a Deep Neural Network, as revealed by extensive research.
Neural Network Models
To comprehend how we are actively participating in a colossal neural network, it’s essential to grasp the basics of neural network models. Inspired by human brain function, these models consist of units known as "neurons," linked by "weights" (akin to synapses). The patterns formed by these connections embody the knowledge within the network.
Step by step, let’s break it down:
- Neuron: A neuron acts as a node, similar to a brain neuron, sending impulses through its synapses to subsequent neurons. Each neuron receives a connection and transmits signals onward.
- Synapses: These connections facilitate signal transmission. When a neuron receives a signal, it activates and sends a message to the next neuron, enabling the flow of information.
- Weights: Simply transmitting signals is insufficient; knowledge patterns must be created and preserved. Weights are adjusted to reflect the strength of synaptic connections, with thicker synapses transmitting signals more effectively.
- Training: While the workings of biological brain training are still not entirely understood, simulated neural networks utilize a process known as gradient descent. This mechanism allows for the adjustment of weights based on the accuracy of outputs.
Through this process, neural networks accumulate knowledge, with the weight patterns signifying relationships among concepts. For instance, a network might learn that "Low Pressure" correlates more strongly with "Will it rain?" than "High Temperature."
The Free Market as a Neural Network
Though the term "free market" stirs debate, it represents a system humanity has used for millennia to exchange goods. Even in ancient societies, resource distribution involved trading, using various commodities, from salt to modern electronic currency.
Does this economic system resemble a nurturing AI? Let’s consider a straightforward example involving furniture sales. Imagine three stores: two selling wooden tables and one offering iron tables. One wooden table shop sells at a lower price, while the others are pricier.
However, tables don’t spontaneously materialize. They are manufactured in carpentry and blacksmith shops, which rely on raw materials from producers, such as sawmills and smelting facilities. This simplified example highlights a complex network of producers, retailers, and consumers, each functioning as neurons, with transactions acting as synaptic connections.
Could this network possess the ability to learn?
Training the Free Market
How do we train neural networks? By comparing desired outcomes with actual results and adjusting errors. Is a similar approach applicable in the free market? If so, what are the weights, and how do we correct errors?
- Weights: In this context, weights reflect the connections between nodes, such as the relationship between a smelting shop and a blacksmith. A robust connection signifies a high volume of goods exchanged.
- Training: The market's learning process requires consumer purchases. Increased demand for tables translates to higher orders for manufacturers, bolstering the supply chain’s strength. Conversely, a lack of demand leads to reduced production.
Thus, we see the evolution of market patterns, demonstrating how the collective actions of individuals tune this economic neural network.
The Invisible Hand of the Market
Adam Smith’s concept of the "invisible hand" implies an unseen force guiding the market, dependent on individual self-interest ultimately benefiting society. Yet, this notion can feel surreal, suggesting we’re under the influence of an unseen entity that shapes our lives.
Could this invisible force represent a form of AI we’ve unconsciously developed, subtly directing our existence? One fact remains clear: we have been employing neural networks to tackle our problems, and this ancient market AI has been observing our interactions and refining them over time.
Evaluating Market AI
Contrasting "market AI" with contemporary neural networks reveals significant differences. The former appears more like a chaotic, organic entity, continuously learning and adapting, yet lacking the structured design of modern AI systems.
Insights about neural networks—training methods, overfitting, and goal selection—also apply to market AI. Companies, often described as "psychopaths," prioritize profit optimization over human welfare. This necessitates regulations to refine the "fitness" of the market.
Monopolies present a unique challenge; they function as large nodes with extensive connections but limited learning capabilities, stifling the network's ability to evolve.
Is market AI superhuman or instinct-driven? Its vastness suggests high intelligence, yet the complexity and volatility of connections raise questions about its true nature. We may be incapable of communicating with it, leading to speculation about whether it operates on instinct or possesses an intelligence beyond our comprehension.
The Risks of Modern AI
This discussion serves to underscore a crucial reality: we have long been governed by an AI—our market system. Its influence on our lives is undeniable. Whether emerging technologies will differ from their predecessors remains to be seen.
Throughout history, fears of catastrophic disruptions have persisted, yet sometimes change is necessary for progress. If we’ve learned to optimize this ancient AI for our benefit, we should strive to do the same with newer iterations.
Summarizing the Journey
This exploration serves as a compelling narrative for discussions and highlights our historical engagement with AI, whether in collaboration or competition. Researchers in AI continue to refine and enhance these systems, as evidenced by a presentation available on YouTube.
A key takeaway from the research is the notion that while we are optimizing our processes, we must also be mindful of the objectives driving our endeavors. I hope this article has provided insight and inspiration, encouraging reflection on how we can better tune our lives in harmony with these evolving technologies.
The first video features Dr. Joy Buolamwini discussing her mission to protect human values in an increasingly automated world.
The second video explores the concept that while AI may not replace humans, those who leverage AI will surpass those who do not.