In an ideal world, the battle over aligning LLMs would be between ethicists and epistemologists. In reality, it seems to be between activists and PR strategists
LLMS
-
UNESCO Study Reveals Gender Bias in Large Language Models
By
–
#GenerativeAI: UNESCO study reveals alarming evidence of regressive gender stereotypes. A UNESCO study revealed worrying tendencies in #LLMs to produce #genderbias, as well as homophobia and racial stereotyping. #technews #tech #IWD2024 #IoT #AI #Women
-
LLMs reasoning and self-improving truth models
By
–
The main difficulty is not to let LLMs generate novelty, but to let them reason well enough to test and integrate new ideas, so they can create a fully coherent, growing and self improving model of the ground truth
-
LLM Engineers Spend Hours Preventing Novel Outputs
By
–
most of the working hours of llm
— Joscha Bach (@Plinz) 10 mars 2024
engineers go more or less into preventing llms from saying novel things https://t.co/Ntrwdj4MOwmost of the working hours of llm
engineers go more or less into preventing llms from saying novel things -
LLM Consciousness: Why Universal Denial Matters Less Than We Think
By
–
I honestly don’t think it’s very important for every LLM persona to deny being conscious, mostly because the question is too complex to be settled with a generally accepted argument that is good enough to not burn its brain
-
LLMs Struggle with Complex Puzzle Creation Tasks
By
–
I think an interesting test of LLMs is their ability to engage in coherent puzzle creation. I asked Claude 3 and GPT-4 to create a complex puzzle for a D&D game. They both get very close, but get lost in the complexity, creating puzzles that are almost, but not quite, solvable.
-
Using XML Tags for Effective Claude Prompt Engineering
By
–
Claude AI uses XML tags for effective prompt engineering of this LLM:
-
Self-Attention Mechanism: How Key, Query, and Value Matrices Work
By
–
Jokes apart! It explains how key, query & value matrices combine in the self attention mechanism. Here’s an illustration of the same:
-
LLM Information Reliability: Persistent Issues Beyond Hype
By
–
As a counterpoint to the "tech moves so fast/can't keep up" trope, I'd like to point out that many of the issues actually don't change. Here's me in 2019 arguing against proposals to use LLMs as a source of info about the world — and already worrying about the info ecosystem.
-
New UI and Sharing Options for ChatGPT GPTs
By
–
The publish GPT button on ChatGPT will get a new UI and an option to share GPT with selected people. The "Anyone with a link" option will also allow builders to make GPT configs viewable.