They took his chair but not his drive

When even your lawyers are excited about what you’re doing, you know you’re truly pushing the boundaries. Back in 2016 at Crypto .com 🦁 we reinvented banking for the first time by re-imagining money and assets as available 24/7. Back in 2022 at Gnosis Pay 🦉 we reinvented banking by re-defining account balances as owned and controlled by the user. And now at @liminalcash 🔆 we are reinventing banking once again by revolutionizing how you interact with money. We’re making the shift from manual tasks on a passive balance ➡️ to intent-based execution on money that understands you and proactively works for you 🧡

In person > remote first

Only gone for a couple days and my monitor gets stolen

Gödel, Escher, Bach — The Eternal Golden Braid. I can’t stop thinking about what it implies for the future of AI. The book is almost 50 years old. But its central question feels increasingly current: How does intelligence emerge from things that are not themselves intelligent? 2/ Gödel approached this through mathematics. He showed that sufficiently powerful formal systems eventually become capable of making statements about themselves. And once a system can refer to itself, something strange happens: there are truths about the system that cannot be proven from within it. 3/ Escher did the same thing visually. Hands drawing each other. Staircases that climb forever only to return to where they started. Every individual step makes sense. The paradox only appears at the level of the whole. 4/ Bach did it through music. A simple theme is copied, inverted, layered and fed back into itself. Simple rules. Recursion. Increasing complexity. Eventually something exists at the higher level that isn’t obvious from any individual component. 5/ Hofstadter calls these structures strange loops. Systems that operate on one level, build representations at another, and eventually loop back around to represent themselves. His much bigger proposition is that consciousness might work this way too. 6/ A neuron isn’t conscious. 86 billion neurons somehow produce something that says: “I am me.” Maybe consciousness isn’t a special substance inside the brain. Maybe the self is a model that becomes sufficiently complex to model itself. 7/ Which raises an uncomfortable question about AI. We keep asking whether machines can become conscious as though consciousness is a feature someone eventually has to code. What if nobody codes it? 8/ Today’s AI systems are acquiring increasingly persistent memory, tool use, planning, reflection and models of their own previous behaviour. None of those things individually imply consciousness. But they create increasingly sophisticated feedback loops between: memory → action → observation → self-model → new action. 9/ Now think of an agent that has a built in memory graph. It remembers you. It remembers itself interacting with you. It acts in the world, observes the consequences, updates its understanding and uses that history to determine what it should do next. That still doesn’t make it conscious. 10/ But it makes the question less ridiculous than it initially sounds. At what point does an agent with persistent memory, recursive self-modelling and the ability to act stop being merely a sequence of independent computations? I don’t know. And perhaps there isn’t a clean threshold. 11/ Gödel, Escher, Bach suggests that we may be looking for consciousness in the wrong place. Not inside any individual neuron. Not inside any individual model invocation. But in the strange loop created when a system can remember, model and repeatedly encounter itself. 12/ The most interesting possibility for the next generation of AI may therefore not be bigger models. It may be continuity. Memory. Identity. Agency. Feedback. A system that doesn’t just process the world, but gradually constructs a model of its own place within it. That is where things get interesting.

Sorry — the whiteboard is strictly reserved for Luke’s head.

Jack you smell nice today