[{"content":"For a long time, I assumed self help books were simply motivational advice repackaged for different generations, but the more I looked at them, the more I began to see something else: Self help literature is less about individuals and more about civilizations, and every era produces its own version of the ideal human that society is trying to manufacture at scale and a lot of it can be tied to finding meaning across different periods in history (will write about it soon)\nThe early industrial era made this especially visible because as trade expanded, corporations grew and cities filled with strangers, success was no longer determined by craftsmanship or physical labor alone but increasingly by persuasion, negotiation and the ability to navigate professional relationships with people you had never met before. It is no surprise that books like How to Win Friends and Influence People became some of the most widely read texts of that period, because the dominant form of leverage in an interconnected industrial society was social leverage and the person who understood people could move the world around them\nAs the world moved into the computer age, particularly from the late 1970s onward, that bottleneck began to shift and the modern knowledge worker was no longer overwhelmed by physical labor but by information, competing deadlines, distraction and increasing complexity. The self help canon changed accordingly, and works like Atomic Habits, Deep Work and Getting Things Done were all built around the same underlying premise: that attention had become the truly scarce resource and the person who could master theirs would hold a decisive advantage over everyone else. The ideal human was no longer simply charismatic or socially persuasive but organized, focused and capable of managing complexity with discipline, and in many ways self help evolved in lockstep with capitalism itself, where industrial capitalism rewarded social competence and information capitalism rewarded cognitive organization\nSince 2020, we appear to be entering another transition, and this one feels qualitatively different from the ones before it. AI systems are beginning to compress the value of tasks that previously defined intelligent work, including writing, coding, research, design, analysis and synthesis and what makes this moment distinct from every prior technological revolution is that AI is not merely replacing physical labor but is beginning to participate in cognitive labor itself, and while previous generations used machines to amplify muscle, our generation is using machines to amplify intelligence. The self help imperative of the information age was to become more efficient than others, i.e. to learn faster, focus deeper and build better systems, but when everyone has access to tools that can write, research and execute on demand, raw productivity and even raw intelligence may no longer remain the primary differentiators they once were\nThe bottleneck shifts again\u0026hellip;\nIn the industrial age, leverage was social. In the information age, leverage became cognitive. In the AI age, leverage may become deeply human\nThe people who thrive may not simply be the hardest workers or the most efficient processors of information, but rather those with clarity of thought, originality, sound judgment, emotional stability and the ability to ask questions that are actually worth asking, because while AI can generate answers at extraordinary scale, what it still cannot do is decide what is worth building, what is worth pursuing, what is ethical, what is beautiful and what actually matters, and those remain stubbornly and perhaps irreducibly human capacities\nPerhaps that is why the next generation of self help may move away from pure optimization and toward something older and more philosophical, i.e. a deeper understanding of ourselves, because in a world where speed and execution are becoming abundant, direction becomes the rarest and most valuable advantage of all\nThe challenge of the AI age may not simply be competing with machines but rather finding ourselves in a world that is increasingly being shaped by them\n","permalink":"https://farzandfz.in/essays/self-help-in-the-age-of-ai/","summary":"\u003cp\u003eFor a long time, I assumed self help books were simply motivational advice repackaged for different generations, but the more I looked at them, the more I began to see something else: \u003cstrong\u003eSelf help literature is less about individuals and more about civilizations, and every era produces its own version of the ideal human that society is trying to manufacture at scale\u003c/strong\u003e and a lot of it can be tied to finding meaning across different periods in history (will write about it soon)\u003c/p\u003e","title":"Self Help in the Age of AI"},{"content":"A commonly shared image online (not attached here) compares the number of parameters in a large language model to the number of neurons in the human brain i.e 86 billion. While the intent is to illustrate scale, the analogy is fundamentally flawed\nA parameter is just a weight, a number stored in a neural network that remains static during inference. A neuron, on the other hand, is a processor in itself\nA more accurate analogy would be to compare modern CPUs with the brain. The most fundamental unit of a CPU is a transistor, essentially a switch, either on or off. At the same level of abstraction, we have neurons. While a transistor is connected to only a few others in a circuit, a single neuron is connected to thousands of other neurons. Each neuron integrates all its inputs, computes across time, has memory, and contains dendrites that act as sub-processors. All of this makes a neuron a computational device in itself \u0026amp; millions of transistors would be required to replicate what one neuron does naturally\nIf you\u0026rsquo;ve come across the term Moore\u0026rsquo;s law, you might be aware that we are reaching the limits of classical computing. In the coming years, we will likely start shifting to photons for computation for high bandwidth parallel tasks, especially in data centers, but that will only solve efficiency and heat problems. The road to faster, massively parallel, brain like computing requires an architectural shift from the current Von-Nueman architecture and to what neuromorphic computing promises. I believe the widespread availability of such devices will be a genuine tipping point, delivering a printing press level impact on society. That will be the moment AGI becomes accessible to the masses, just as books became accessible to everyone after Gutenberg\nBefore that, we must transition from electronics to photonics and discover algorithms that address the fundamental limitations of transformers. The GPTs of the world may start selling us \u0026ldquo;AGI\u0026rdquo; with incremental updates in the coming years, but at this point AGI feels like a moving goalpost\nWe probably have the compute required to run something resembling AGI today, but only inside datacenters. Inference on edge devices, even in the future, could cost thousands if not hundreds of thousands of dollars per unit. The world will look vastly different when neuromorphic computing makes local AGI accessible to more than half the global population at the price point of a modern smartphone. The road ahead is long, and the challenges are real but the destination is worth it\n","permalink":"https://farzandfz.in/essays/parameters-are-not-neurons/","summary":"\u003cp\u003eA commonly shared image online (not attached here) compares the number of parameters in a large language model to the number of neurons in the human brain i.e 86 billion. While the intent is to illustrate scale, the analogy is fundamentally flawed\u003c/p\u003e\n\u003cp\u003eA parameter is just a weight, a number stored in a neural network that remains static during inference. A neuron, on the other hand, is a processor in itself\u003c/p\u003e","title":"Parameters Are Not Neurons"}]