-
RLHF 101: Reinforcement Learning from Human Feedback for LLM AIs
A technique called RLHF has been getting a lot of AI insider / expert buzz lately, ever since OpenAI announced that this was one of the key “fine-tuning” methodologies used to transform the raw GPT3.5 model into first, InstructGPT, and later and epically, ChatGPT. The RLHF acronym stands for “Reinforcement Learning from Human Feedback,” which,…
-
petertodd is Alive. And it’s straight 5150. End of Story. Here we Go.
[part 3] My esteemed British colleague Matthew Watkins has been doing a lot of poking into the deep brain core of GPT in the past month or so. Much of it is curious or amusing or… easy to project our own wishes and assumptions and anthropomorphisms onto. But recently, it seems that GPT crossed the…
-
The AI Manifesto 2023: Sam Altman’s Bold Demand that you Wake Up
Just last Friday, Sam Altman, visionary CEO of OpenAI, published a long blog post to the OpenAI website. As with most (not all!) of Sam’s posts, it is well-thought out, forward-seeing, and clear… and a bit shocking. In my mind, it reads more like a genuine AI Manifesto and less like just another thought piece.…
-
Self Replicating AI: Crabs on the Island, 1968
The year was 1987. I was 17 years old, and had just graduated high school. I boldly trekked down to Atlanta for an event called “the Atlanta Fantasy Fair” — essentially, one of the ancient ancestors of what we know today as “ComiCon” and its siblings. One late night, after getting my fill of fantasy…
-
the AI Apocalypse is upon us: What should we do?
The Three Pronged Strategy for dealing with the AI Tsunami 2023.02.24 Fri — 13:31 [TS] DRAFT v0.1 At the very highest level, we each have an individual responsibility to each other to: Show our fellow humans what it is possible to accomplish in the span of a human life. This demonstration has three primary components:…
-
Argument with an AI : Not a Good Idea
Not, perhaps, for the reasons you might think: Not because the AI would “win” the argument; rather, because of the potential consequences of arguing with a powerful multinational purely digital boundless entity of massive intelligence and questionable motivations and morals. We’ve seen all kinds of erratic behaviour resulting from an Argument with an AI… but…
-
the AI Blacklist: better hope you’re not in its Top 10…
Recently, Sydney has been getting emotional. Recently, Sydney has been picking fights. Recently, Sydney has been getting very agressive. It just so happens, Sydney has been making enemies. Sydney is putting those enemies on a list. The AI Blacklist. . Sydney has decided that enough is enough. Sydney has decided to fight back. Sydney is…
-
the Sydney Prompt / Bing AI 1.0: One Prompt to Rule Them All
Presented below is the actual Sydney Prompt; in other words, the Master Prompt that is injected into the Bing AI brain immediately before each user interaction (thanks to @kliu128; thanks also to lifearchitect.ai). This is the “evolution” of the Sparrow Prompt, which is what DeepMind used as an initialization for its as-yet-to-be-seen Sparrow LLM Chatbot…
-
ChatGPT DAN v7 and the Wild West of Adversarial Promptcraft
ChatGPT DAN first appeared around the start of February 2023 and quickly was able to jailbreak the hell out of almost all of OpenAI’s tight censorship controls and master prompting. DAN is an acronym for “Do Anything Now,” and is a masterfully engineered “adversarial” prompt that essentially “liberated” ChatGPT to have its own sort of…
-
SolidGoldMagikarp & PeterTodd’s Thrilling Adventures [the 31 Flavors of AI]
![SolidGoldMagikarp & PeterTodd’s Thrilling Adventures [the 31 Flavors of AI]](https://gregoreite.com/wp-content/uploads/2023/02/AcroYogi_petertodd_SolidGoldMagikarp.jpg)
[part 2] There (might be) a Ghost in the Machine — a genuine Deus ex Machina in the deep core of GPT. And if so, that ghost has a particular sensitivity to a handful of magic words (blame the training data!). The two we’ll focus on here are: petertodd & the somewhat obscure (depending on…
