Mindscapes & Metalogues 🕘 история названий (1)

Consuming information Fast and Slow…. After working extensively with AI agents over the past two weeks, I've noticed two patterns for consuming and dealing with information. The first is “fast”, to just bridge some knowledge gaps; the second is “slow”, to allow my brain to rewire. When I need a specific, real-time answer to clarify a detail, resolve an issue, or gain a high-level overview, I ask AI to summarise and highlight the essentials. That's very efficient and direct. AI is the king here. Conversely, when I want to deeply internalise certain ideas and allow my brain to be rewired, I must interact with the content slowly. This is a fundamentally different approach. Engaging with a book or a long podcast is like inviting another person’s perspective to interact with your neural network. We are the sum of people, books, movies, cities, and environments we encounter; they all shape our “mindscapes”. Therefore, to cultivate taste, authenticity, and your identity, slow consumption is essential. A recent example: I was reading a book where the author briefly referenced the myth of Aphrodite and her dual identity. The author likely assumed his readers are all highly educated and cultured. Not being in this category, I turned to AI to quickly bridge the gap. I didn’t need a deep academic analysis of Greek mythology; I simply needed a targeted piece of information to continue with the text. However, I would never ask AI to summarise the book itself. I am not reading it for mere information, but to allow my brain to encounter and interact with the author’s unique way of thinking and worldview.

We might soon need a 'gym for the brain' with mathematical problems to solve, just to keep our brains from fully atrophying.

Uncle Bob confirms.

Surprisingly, the Agile principle of 'customer collaboration over contract negotiation' applies perfectly to interacting with AI agents. They may miss details in the initial specification, but the ongoing conversation drives the necessary fixes.

“AGI is the ability to figure things out.” “A human who can figure things out has some baseline knowledge, the ability to reason over that knowledge, and the ability to iterate their way to the answer.  An AI that can figure things out has some baseline knowledge (pre-training), the ability to reason over that knowledge (inference-time compute), and the ability to iterate its way to the answer (long-horizon agents). The first ingredient (knowledge / pre-training) is what fueled the original ChatGPT moment in 2022. The second (reasoning / inference-time compute) came with the release of o1 in late 2024. The third (iteration / long-horizon agents) came in the last few weeks with Claude Code and other coding agents crossing a capability threshold.” "If there’s one exponential curve to bet on, it’s the performance of long-horizon agents. METR has been meticulously tracking AI’s ability to complete long-horizon tasks. The rate of progress is exponential, doubling every ~7 months. If we trace out the exponential, agents should be able to work reliably to complete tasks that take human experts a full day by 2028, a full year by 2034, and a full century by 2037. " Source: https://sequoiacap.com/article/2026-this-is-agi/

AGI is born! Truly it is born! Optus 4.5 is a breakthrough. Just today, I completed two tasks that would otherwise take me a couple of days: 1. Created detailed requirements for a product feature (M size). 2.Analyzed text transcripts of 30 customer calls, detailing use cases, major questions, and objections. And that’s just the beginning. In 4-6 months, there will be another model. What are we all going to do with all this power in our hands?

The mismatch between "corporate propaganda" and reality is astonishing. All companies claim that their people are their most valuable asset. Yet, when the time comes to trim the budget, employees are the first item slashed to keep costs low. What’s worse is that this behavior is pre-determined by a glitch in modern accounting: under current rules, employees are classified as expenses to be minimized rather than assets to be grown. If we moved "people costs" from the expense column to the asset column on balance sheets (treating them like true capital assets) we might finally encourage companies to be more human-oriented. https://thesystemsthinker.com/if-people-are-assets-why-do-we-treat-them-like-expenses/

Curiosity is becoming one of the most valuable skills. This is what differentiates us from AI.

Product teams will prevail.

Привет! В глубинах YouTube наткнулся на фантастический курс о сложном мире от Георгия Сатарова. Кроссдисциплинарность зашкаливает: физика, математика, информатика, биология, нейробиология, астрономия, социология, политология, психология. И все это про хаос, эфолюцию и способы развития в рамках постоянно меняющейся среды (aka институты хаоса). Три наиболее интересных тезиса: 1. Хаос и порядок порождают друг друга. Граница между ними практически не различима: простые детерминированные алгоритмы порождают хаос; хаос разбивается на простые «осязаемые» компоненты. 2. [эволюция] Селекция и закрепление не скоординированы. Ни одна система не может управлять своей эволюцией: в силу неопределённости будущего никто не знает, какие характеристики позволят системе сохранить себя. Желательно иметь «излишек» характеристик, которые позволят быстро адаптироваться к изменившейся среде. 3. Система способная к обновлению получает шанс (!) на продолжение жизни. И для ценителей: произведение Баха о хаосе и порядке. P.S. Поскольку курс рассчитан на максимально широкую аудиторию, динамика достаточно медленная. Рекомендую слушать лекции на x1.5-x2