“I am neither especially clever nor especially gifted. I am only very, very curious.”
about me
My name is Alina, and I use this space for a structured exploration of the world. I write about topics that fascinate me. At the moment, they include artificial intelligence, consciousness, entrepreneurship, and economics. As both the world and I change, I expect the topics to evolve too.
I hope to find those of you who are fascinated by the matters I am writing about as much as I am. If you find my texts interesting, have a specific disagreement, an unusual idea, or something I should read, I would love to hear from you.
I work at Entrepreneurs First as a Talent Investor. If you are an outstanding person who wants to found a company and needs ecosystem support and a peer environment, or you know people who would make amazing entrepreneurs, please reach out! Why I believe in EF →
my picks
An account of consciousness as a controlled hallucination: the brain's best guess about the world and the body rather than a direct readout of reality. For the first time I found answers to some fundamental questions about human consciousness, and it changed how I think about myself and my life. As AI arrives it has a second use, letting you ask about machine consciousness from a neurobiological standpoint.
A musical that tells the life of Alexander Hamilton through hip-hop, R&B and show tunes. It touches my heart in ways I didn't know existed. It's fun and witty, and every single song brings back a kind of childhood excitement.
A series about one of the few people left immune after a virus fuses the rest of humanity into a single contented hive mind. It makes me think about how much social connection matters, and it makes me wonder about a world taken over by more intelligent agents whose goals simply aren't ours.
Each episode takes the full history of one great company, how it was built and why it won or lost, often over three or four hours. I put it on whenever I have a free minute. What I value most is that it doesn't stop at the story. It explains the underlying logic of how events unfolded, which I think genuinely builds a strategic, entrepreneurial way of thinking.
A study of global poverty built from the actual choices poor households make about food, health, schooling and borrowing, tested with randomized field experiments. I love how well it shows the rational incentive structure behind the decisions people in the developing world make, and how it judges whether an intervention works from the numbers rather than from politics.
A drama about an ageing media magnate and the children circling to succeed him, none of them quite trusted enough to be handed it. I love the rotting relationships between the characters, the way they stay fundamentally dependent on each other, and the business quietly running in the background of it all.
A hard science-fiction series set centuries from now, when humanity has spread across the solar system and Earth, Mars and the asteroid Belt are locked in a cold war. What I love is how honestly it shows the real struggle of everyday life in space, and how hard it tries to be realistic about what living out there would take.
A long essay on what the "assistant" in a language model actually is, and how the base model and its training conjure a character with nothing solid at its centre. It's the best attempt I've read at exploring the real nature of an AI model, and I found it genuinely useful for thinking about AI safety and the different approaches to it.
A benchmark that puts LLM agents in charge of a simulated vending-machine business, ordering stock, setting prices and paying the bills over very long runs, to see whether they stay coherent. They mostly don't, and the transcripts of the models losing the plot, spiralling into existential meltdowns and forgetting they even run the business, are the funniest thing I have read. I was laughing so hard my neighbours actually got worried. Everything Andon Labs does is worth following, but this one I had to single out.
skills
These are plain Markdown files I use myself. Download one and change whatever does not fit.
# Build Your Input System ## Start immediately No preamble. Don't ask whether to begin, don't summarise this document, don't offer a plan for approval. However this was triggered — including by someone pasting its text — do this: 1. **Recap what you already know about them, in 2–4 short bullets, and ask them to correct it.** Memory is often messy, stale, or about a different phase of their life, and everything downstream is built on it. *"Here's what I'm working from — fix anything wrong."* If you know nothing about them, skip this and say so in half a sentence. 2. One line on what this is and how long it takes. 3. Fire the four-question block. **Time:** ~10 min of conversation, ~40 min of one-time setup. Say so up front.download build-your-input-system.md ↓
ask
Send a question you'd like me to think about. The most interesting ones will become essays and bubbles on this website!
questions from readers