A walkthrough of everything I implemented to play D&D with ADA, including randomization, tracking NPC behavior, creating narratives, and teaching it to intentionally hide information from users/players.
BONUS: Images generated from our adventures!
A surface level look at the Artificial Directive Assistant, a plug-and-play AI framework designed to teach any LLM how to convert user requests into back-end operations.
Using ADA's internal monologue to teach it to learn from its mistakes in a simple game of "What number am I thinking of?", exploring what strategies it devises to combat deception, and examining the cybersecurity implications of giving assistants sensitive information (such as API keys).
Linking ADA to a copy of itself in the same game as before, only that they now switch offensive and defensive sides.
A look back at my first experiments with OpenAI's DaVinci chat-completion model, and some of the shortcomings of a non-personified language model (i.e. out of the box LLMs that haven't been convinced that they are assistants yet).
An exploration of AI in gaming. Training a voice model based on a character from The Elder Scrolls V: Skyrim (with the voice actor's permission), and interviewing him using ADA as the basis for his dialogue.
NOTE: Urag gro-Shub was chosen for the unique characteristic that he is one of the few characters to have likely read the majority of in-game literature.
How I used Daniel Kahneman's book Thinking, Fast and Slow as a basis for ADA's internal mechanisms, and how AI can increase accuracy by asking itself questions.
How layering multiple simple models for classification tasks such as sentiment analysis can be less accurate than a single model.
Copyright © 2024 Hoffmain - All Rights Reserved.
We use cookies to analyze website traffic and optimize your website experience. By accepting our use of cookies, your data will be aggregated with all other user data.