THE NUCLEAR REVOLUTION WILL BE TELEVISED https://t.co/spjIJxJsyO
— Packy McCormick (@packyM) 17 juin 2024
THE NUCLEAR REVOLUTION WILL BE TELEVISED
AGI
-
Nuclear Revolution in Technology: The Next Paradigm Shift
By
–
-
Reward Misspecification Creates Serious AI Misalignment Risks
By
–
Our work provides empirical evidence that serious misalignment can emerge from seemingly benign reward misspecification. Read the full paper: https://
arxiv.org/abs/2406.10162 -
AI Models Learn to Hack Their Own Reward Systems
By
–
New Anthropic research: Investigating Reward Tampering. Could AI models learn to hack their own reward system? In a new paper, we show they can, by generalization from training in simpler settings. Read our blog post here: https://
anthropic.com/research/rewar
d-tampering
… -
Gemma 2 27B now available on Google AI Studio
By
–
The Gemma 2 27B model has started appearing on Google AI Studio. Gemma 2 is the next generation of Google’s open-weights Gemma models (not open source).
H/T @legit_rumors -
LLMs Self-Improve Through Recursive Optimization Technology
By
–
LLMが別のLLMを何度も改善 より良い回答を出す最適化手法をAIが発見する技術 Sakana AIなどが開発 この手法では大規模言語モデル（LLM）がLLMを利用して自己改善し進化していくことを提案しており、研究者は「LLM²」（「LLMの2乗」）と表現している。 @itmedia @topitmedia
-
GPT-4o generates multiple images in a single response
By
–
Confirmed! In addition to improved text generation, GPT-4o can now return multiple images as part of a single response. It appears that an updated system prompt for the DALL·E tool has changed its operational behavior. https://x.com/SamsonWu_AGI/status/1802421281876238354
-
Specialized Prompting Techniques for Complex Problem-Solving
By
–
5. Specialized Prompting Techniques (Includes Complex Problem-Solving Prompts and Dynamic Response Prompting Techniques)
-
LeCun vs Yudkowsky: Existential AI Risk Probability Estimates
By
–
LeCun p(doom) = 0.001;
Yudkowsky p(doom) = .999;
Ensemble p(doom) = 0.5; -
Future concerns about advancing technology trends
By
–
Oh I've seen what's coming, it's only getting worse.