Recent content on Paul Payne
Software Needs a Home # You may have hear the term “ Vibe Coding ”. It just means writing software with AI assistants. You can ask most assistants to write code for you presently and they’ll do a fair job of it. There’s some serious work to be done , still, in getting them to write good software, but, really it’s already good enough form many types of software.
The Dream # I want to talk about you and me, and our, and our society’s relationship with technology. But allow me to start with a touch of history. In 1936, mathematician Alan Turing conceived of theoretical machines that could be generally configured and reconfigured to operate in nearly infinite ways. He called these “Universal Turing Machines” because he imagined them being…
A lot of my recent thoughts about coding with agents lately (like this , and this , and this ) have been circling a specific topic… design. Coding agents are now good enough to replace all of my manual coding tasks, I haven’t written code myself for months (and it’s not because I don’t know how), but we’re still figuring out how to make them better at the more…
Don’t distract the engineers # In January, I wrote about how best-practices make software development agents work better. Now, six weeks later, we’re in a position where we’re not just talking about individual developers with a coding agent, but teams of developers with fleets of agents. So, it’s time to update my thinking.
We saw a massive sell-off of software stocks last week, partially in response to Anthropic’s claim that 100% of their code is now written by AI and the release of a couple of new products for lawyers and office workers. I’ve been doing this for a few years now, so let me provide you with a bit of insight into how this actually works.
There’s a charged debate happening right now about whether or not coding agents are worth a damn. I’m going on the record and saying, yes, yes they are. There are some people who just tried coding assistants at various stages of development and didn’t like their output. I hear often is that AI code is slop; AI can’t produce anything beyond toy software; that when you reach…
I’m calling it. My entire 30 year career in technology is a historical artifact. My current work with AI has convinced me that, in a few short years, humanity will look back and think how funny it was that people once actually programmed machines.
At the beginning of the year I became increasingly concerned with the state of civil society here in the U.S. and around the world. Polarized online discourse, mental health crises, centralized control, challenges to privacy, continued ecosystem degradation, run-away growth, shifting political landscapes, attacks on civil liberties, and an overall disappearance of civil society as a whole in the…
For now, the intelligence in our systems remains distinctly human. Eighteen months ago, I wrote about building thinking machines using LLMs , driven by curiosity about whether these systems could help us understand human consciousness. What I’ve learned building LLM applications since then fundamentally changes that original question.
Current work # I am a Principal Research Engineer at Microsoft in the Office of the CTO . My current work is building AI systems by augmenting generative models with systems and cognitive architectures. We explore what we can do with LLMs to unlock their utility to a larger range of applications and spend our time building functional prototypes and sharing our learnings across Microsoft, academia,…
How the 2025 presidential inauguration reflected a breach of social contract by big tech and one person’s first steps at reclaiming their digital autonomy.
“an abstract illustration of consciousness studies, timeline, gradient” –Midjourney Early Investigations # In our investigations of creating machine minds, let’s start with a quick survey of previous thinking on the topic. Machines have only relatively recently come on the scene, but our investigations of mind and consciousness trace back thousands of years. This article is a…
“a beautiful photo of a memory system, neurons, warm and comforting” –Midjourney Imagine for a moment life without memory. Each moment is a fresh experience of the world. Where are you? There would be no concept of narrative, how you got here. What is this place? There would be no firm semantics, what is here? Who is the other person in the room? What is a person? What is other?…
Let’s talk about the suitability of LLMs (Large Language Models) for exploring thinking machines. Are they all we need? Where do they fall short? What areas are they especially suited to? But first, a bit of LLMs background to get a better understanding of them. A Brief History of LLMs # LLMs were developed from the academic field of natural language processing (NLP), which studies the…
“surrealist painting of the creation of an autonomous agent, machinery, consciousness” -Midjourney The simplest agent one can build might be: def hi (): print ( “ hello ” ) This exhibits the behavior one would expect of an agent in the world. In response to an external stimulus, a function call, it reacts, printing “hello” to the console from which you called it.
Cover: “photo of a boy on the floor reading an old encyclopedia, sunlight, scattered toys, 80s” –Midjourney I hope these articles will open collaboration with others of you who share my interests. And, if we’re going to be working together , then you should know something about me. To that end, before we dive into the philosophy, theory, and technology, I want to give a bit…
“A beautiful technical drawing of artificial cognition architecture. muted pastel lines with neon accents on a white background…” -Midjourney This newsletter will chronicle my research into understanding consciousness, specifically through the lens of crafting it using the latest advancements in artificial intelligence. I am Paul Payne, a Principal Research Engineer in the…
“a blueprint of a synthetic memory machine” –Midjourney ChatGPT is Forgetful # ChatGPT is forgetful. More specifically, it doesn’t remember anything. It responds to your messages, so it feels like it remembers you. It even uses some of the history of your chat as it crafts new responses, so it can be a convincing charade, but if you’ve ever tried to have a long chat with ChatGPT,…