WRITINGS

• Al as a Digital Mirror, the Illusion of Depth, and the Difference Between Processing Information and Understanding It:
Douglas Adams said that art holds the mirror up to nature. Al has the same function: it mirrors human intelligence. It's not intelligent. It doesn't think, understand, or learn, certainly not in the context of human cognition, consciousness, or understanding. Rather than thinking, it simulates thinking through statistical analysis (finding correlations in massive datasets), pattern recognition (remixing existing information in its training data) and data processing (executing complex, recursive algorithms).
It's an infinity mirror, a physical, real-world example of the Droste effect, which is a type of recursive, picture-in-picture phenomenon. But, a "mise en abyme" reflection doesn't equate to intelligence.
A mirror or a similar reflective system like Al, acts as conduit for data-capturing and reflecting, but it doesn't possess the internal awareness required for intelligence. It reflects, amplifies and recursively iterates on human cognition and data. Al is loosely inspired by biological processes and neural networks in the human brain. It can perform complex calculations, identify patterns, and simulate human-like reasoning, but it operates fundamentally differently than the human brain. Because it produces humanlike output, it's tempting to attribute intelligence to it. Al is a reflection of human intelligence rather than a creator of its own. In essence, art and Al mirrors are instruments that reveal the world, but it's the human brain that interprets that revelation.

• Organic Origin, Functional Parallels, Embodiment, Consciousness, AGI & Superintelligence:
Shifting the focus from simply scaling up models to ensuring "safe Superintelligence" doesn't hold much weight, as most experts cannot agree on what Artificial General Intelligence and Superintelligence are, nor can they answer the question:"What does it mean to be conscious?". Rapid advancements in neurotechnology and Al are outpacing our understanding of consciousness. They argue that identifying consciousness in machines and brain organoids could "force society to re-think responsibility, rights, moral boundaries, and empathy". But, should we "force" society to do anything? Is that the right approach?
I know this much: doom-mongering works. And it sells books. When people hear or read the phrase, "Al is an existential threat" dozens of times per day, they start believing it. Hype, fear, rage bait, ambiguity, and uncertainty.

Humans are wired to prioritise negative, threatening information over positive or neutral information (the negativity bias). Propaganda is a potent, deceptive tool designed to sway public opinion and emotions to serve specific commercial, political, or ideological agendas. It shapes perceptions and drives actions, frequently eroding trust and fuelling social polarisation. I try to combat pessimism, alarmism, doomism and similar exaggerated narratives because they distort reality and often hinder rational solutions.

-Machines cannot develop "human-level intelligence" because they don't have human cognitive architecture. They cannot genuinely think, understand, learn, remember or feel, at all. A simple yet critical distinction between mimicking these human traits and actually experiencing them. AGI and ASI/ Superintelligence are often conflated. I disagree that humans are paradigm examples of general intelligence. In fact, I think that general intelligence doesn't exist at all, even in humans. Roger Penrose argues that human consciousness is noncomputational and cannot be replicated by Al or algorithms alone. I concur. He proposes that our consciousness stems from nonalgorithmic physics, potentially involving quantum processes (qubits) rather than just neural network computations. His Orch-OR theory suggests that quantum coherence occurs within structures called microtubules inside neurons. This is where consciousness is generated, Penrose claims, as these structures are quantum devices that orchestrate awareness. Al cannot become conscious just by becoming more complex or powerful; it simply simulates intelligence and processes information without truly experiencing it. I disagree with the researchers in Al who believe that simulation is enough for intelligence and that functional equivalence to the brain will eventually bring about consciousness.

• Al Is a Tool Built upon Code and Data: It's a sophisticated tool for calculation and prediction, but it's still just a tool. It can be a creative multiplier/amplifier but it requires creativity and a clear vision on my part. It's not some prepubescent version of Agent Smith that is rapidly developing, becoming a "superintelligent" and preparing to terminate human simulacra :) Al is a tool fundamentally built upon computer programs, algorithms, and data structures. While it's often viewed as a revolutionary technology, it's a branch of computer science focused on developing software or machines that simulate human cognitive functions, such as learning, reasoning, problem-solving, and perception.

1. Al systems are not sentient but are highly complex, specialised software applications.

• Algorithms: Al uses algorithms to process data, find patterns, and make predictions.

• Machine Learning (ML): This is a core subfield of Al where algorithms process data to identify patterns and make decisions without being explicitly programmed for every specific scenario.

• Deep Learning (DL): A subset of ML, DL uses multi-layered neural networks (modelled after the human brain) to process complex data.

● Programming Languages: Al systems are built using various programming languages, including Python, LISP, Java, C++, and others.

2. Traditional Programming vs Al: Instead of being told how to solve a problem, the Al is "trained" on vast datasets to identify patterns. The Al then develops its own internal "logic" to make predictions or decisions.

3. How Al Tools Operate: Al acts as a tool that enables machines to perform tasks usually requiring human intelligence. Examples include:

• Generative Al (e.g., ChatGPT, Claude): These models use Natural Language Processing (NLP) to generate human-like text by identifying patterns in massive datasets of text.

• Computer Vision: Algorithms allow computers to interpret visual information from images and videos.

• Recommendation Engines: Algorithms that analyze user behavior to suggest content, such as on Netflix or Amazon.

4. Limitations: They cannot adapt to completely new, unseen situations (lacking common sense), and they are limited by the quality of their training data. They are, at their core, sophisticated pattern-recognition tools operating on statistical correlations.

Summary Thought: Al is a powerful, specialised tool for calculation and prediction rather than an intelligent entity. However, it can simulate intelligence to such degree that it often appears to be thinking, which is a major area of ongoing debate and research.