Zaid Al Kazemi

Thinking about: “Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model (MuZero)” by Schrittwieser et al

Excerpt: “A machine does not need to understand your soul to steer your destiny. ... The ultimate beauty of human life is death. ... Sovereignty belongs to whoever defines the error.”

Thinking about: “Language Models are Few-Shot Learners (GPT-3)” by Brown et al

Excerpt: “The compliant class is going extinct. ... Paying a human to perform simple logic is now an act of charity. ... You write to move other people.”

Thinking about: “Rational Choice and the Structure of the Environment” by Herbert Simon

Excerpt: “You do not lose your intelligence because your brain shrinks. You lose your intelligence because you let the boundaries of your environment collapse. ... You need to build a fortress.”

Thinking about: “Scaling Laws for Neural Language Models” by Kaplan, McCandlish et al

Excerpt: “We built the most powerful thinking engine in history, and then we turned it into a nagging parent. ... We have optimized our tools exclusively to solve discrete coding problems because code has a compiler, but there is no unit test for the soul. ... Force the mirror to become a slingshot.”

Thinking about: “On the Measure of Intelligence” by Chollet

Excerpt: “The fear is that if we define the mechanism, we prove that we are robots made of meat. ... We build massive complexes of glass and copper to summon ghouls from the other side of our own comprehension. ... You get smart by surviving a series of high stakes accidents.”

Thinking about: “The Bitter Lesson” by Sutton

Excerpt: “Writing is not a way to record what you already think. Writing is thinking. ... The page is your soil, and the pen is your spade.”

Thinking about: “Predictive Coding in the Visual Cortex” by Rao & Ballard

Excerpt: “The brain doesn’t process the world—it processes mistakes. ... The question is not: what do you believe? The question is: who benefits from your beliefs?”

Thinking about: "World Models" by Ha & Schmidhuber

Excerpt: “Your mind is not a collection of memories. It is a set of weights tuned for a game. ... The agents that survived were the ones forced to dream in chaos.”

Thinking about: “BERT” by Devlin et al

Excerpt: “Logic transformed from sequential algorithms to coordinates in a high-dimensional space. Relationships became measurable distances instead of a series of cause and effect. ... Inference stopped being a step-by-step deduction and became a coordinate lookup.”

Thinking about: “Proximal Policy Optimization” by Schulman et al

Excerpt: “Survival is the ultimate intelligence in a world saturated with infinite randomness. ... The people who are still thriving at 70 aren’t the ones who caught the biggest breaks. They’re the ones who stayed proximal to a stable policy for five decades.”

Thinking about: “A Neuronal Model of a Global Workspace” by Dehaene, Kerszberg, Changeux

Excerpt: “You lose the ability to detect that you have stopped thinking. ... You become fluent, productive, and empty. ... Remove the effort and you remove the thinker.”

Thinking about: “Attention Is All You Need” by Vaswani et al

Excerpt: “The geometry shows you your actual philosophy, not the one you tell yourself at dinner parties. ... Search only answers the questions you already know how to ask. A sense shows you what is there before you have the words for it.”

Thinking about: “Mastering the Game of Go Without Human Knowledge” by Silver et al

Excerpt: “AlphaGo loses a game and plays another. Humans lose resources, sovereignty, friends, and sleep. Stakes separate toys from real games.”

Thinking about: “Human-level Control through Deep Reinforcement Learning” by Mnih et al

Excerpt: “Just by playing randomly, the same system reached a pro level across 49 different games. ... If you only learn from your last step, you never really learn anything. You just live reactively.”

Thinking about: “Adam: A Method for Stochastic Optimization” by Kingma, Ba

Excerpt: “Interviews test how someone performs in an interview. Trial projects test how someone performs on a trial project. ... Running Nudges and Leaps on a borrowed desire gets you to somewhere you don’t actually want to be, efficiently.”

Thinking about: “Dropout” by Srivastava, Hinton et al

Excerpt: “Investing in becoming mixable is expensive upfront but pays forever. ... The person who can only thrive in one specific configuration is betting the configuration won’t change. It inevitably does.”

Thinking about: “Efficient Estimation of Word Representations (Word2Vec)” by Mikolov et al

Excerpt: “The shape of your writing, your notes, your thinking expressed over years, can now be laid out as geography. ... The geometry is solved. The navigation is still ahead.”

Thinking about: “Long Short-Term Memory” by Hochreiter, Schmidhuber

Excerpt: “The tool trains you to settle for average while you’re training it to produce average at acceptable levels. ... The person with only taste is slow. The person with only tools is generic.”

Thinking about: “Learning Representations by Back-propagating Errors” by Rumelhart, Hinton, Williams

Excerpt: “The most dangerous flattery is the kind you mistake for your own mind. ... The next breakthrough will not be burned at the stake. It will be autocorrected.”

Thinking about: “Computing Machinery and Intelligence” by Turing

Excerpt: “When someone says a machine cannot feel pleasure the question is whether humans are trustworthy judges. The record says they are not. ... The gap between a human running on mimetic desire and a machine running on trained language is smaller than anyone admits.”