195 mentions · 155 chunks · 67 episodes
...er than any individual player at any layer can make, which removes competition. ChatGPT is vertically integrating the model, the memory, and the application into one.
...ting mainstream attention; The Wall Street Journal: He Had Dangerous Delusions. ChatGPT Admitted It Made Them Worse. Although the "ChatGPT admitted" frame is misleading, these models can't introspect like that.
...relationship, send them this guide from Less Wrong: "So You Think You've Awoken ChatGPT"
...OADS of fanfic: https://scp-wiki.wikidot.com/scp-series" As a friend observed: "ChatGPT is effectively the memetic equivalent of Gain-of-Function research on viruses but without any containment whatsoever" I've interacted with folks who ...
...ctions of an OpenAI engineer say that at OpenAI "Chat runs really deep." "Since ChatGPT took off, a lot of the codebase is structured around the idea of chat messages and conversations. These primitives are so baked at this point, you sh...
...oes up. Whereas the knowledge graph never gets easier. The 'knowledge graph' in ChatGPT is currently smooshed within the ChatGPT history, but in an ad hoc / messy way–and in a way hidden from the user! Without the user being able to see ...
...er context. The longer you've used it, the more data you've likely accumulated. ChatGPT had a few months pass before it flipped the memory feature on, which led to the "you are insecure about…" embarrassing reveal to friends. People forg...
ChatGPT's Agents feature feels fundamentally reckless to me. Their approach to prompt injection is basically: tell the model to really, really, focus on not ...
...tic about LLMs potential for humanity, and pessimistic about the slippery slope ChatGPT is on. Not where ChatGPT is, but the drain it will circle. Every hyperscale product over time is forced to fall down the slippery slope of engagement...
ChatGPT's answers feel to me like being served up a personalized Axios article. More formatting than substance. Punchy and "simple" while obscuring nuance. P...
...s a big determinant in how good the results are. This is one of the reasons the ChatGPT dossier memory feels off to me. You can't inspect the context to say, "include this, not that". The chat is append only, not coactive. You can imagin...
...th LLMs: People Are Being Involuntarily Committed, Jailed After Spiraling Into "ChatGPT Psychosis" Examples in this Ezra Klein interview.
... the web bigger. "As you think about the evolution of AI and the form factor of ChatGPT … we might still be in Yahoo directory."
...ol to help you think less? This is what is called cognitive debt. Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task
ChatGPT tops out as a Chatbot. Is chat really all there is? Chat is a feature, not a paradigm. Which would you choose between 1) a product that's only a Chat...
... play. Gemini generally does well. Claude refuses to lie, and thus loses often. ChatGPT o3 often wins because it is very happy to betray its collaborators. The things that LLMs are selected for and trained for matter, especially when you...
I used ChatGPt to identify an odd piece of hardware in my post-war AirBnB in Manhattan this week. That could have never worked in Google Lens, but it worked great i...
Isn't it kind of weird when ChatGPT uses the first person in its answers? When you talk to a person you'd never say "I'm just talking to a collection of synapses." We can all agree that...
...ne's actions didn't change given they were being recorded. Do we try to impress ChatGPT? Before the memories feature I didn't care. But now I do. An existential risk: human addiction to "therapy" that feels like growth but actually isn't...
This week ChatGPT gaslit me about its memory feature. I started off asking what it knew about me. I had to really twist its arm to get it to give me most of the things...