Arguably, there is a difference between a simulation (recreation of a causal structure on a different substrate) and a simulacrum (recreating observables without recreating the causal structure)
SAFETY
-
AI Consciousness Binary State Self Awareness Variation
By
–
My own consciousness is quite binary (I am either conscious or unconscious), even though my content and degree of self awareness varies.
-
Scale AI Discusses Red Teaming and Responsible AI at GTC24
By
–
Join this session at #GTC24 on March 19 as @scale_AI discusses topics like red teaming, reinforcement learning from human feedback, and frameworks for responsible AI to ensure that fine-tuned models behave as intended. Register now. https://t.co/I8xsl0bCF7 pic.twitter.com/SCmhN7Df0Y
— NVIDIA AI (@NVIDIAAI) 5 mars 2024Join this session at #GTC24 on March 19 as @scale_AI discusses topics like red teaming, reinforcement learning from human feedback, and frameworks for responsible AI to ensure that fine-tuned models behave as intended. Register now. https://
nvda.ws/48GWLni -
LLM Hallucinations and Prompt-Hacking Performance Analysis
By
–
The tweet you quoted is wrong; I think the person clarifies right in the last tweet.
Its comparing apples to oranges — comparing OOB performance to prompt-hacking performance. Regardless — there's occasional hallucinations and weirdness in AINews because its LLM-Summarized -
Anthropomorphization in AI model learning descriptions raises concerns
By
–
Very very cool work and don't want to take away anything from the magnitude of your achievements but the level of anthropomorphization of your tweet makes me uncomfortable. Like for example, I feel like it's misleading to say "help these models learn more and more" and refer to
-
Racist AI datasets in anthropomorphic tools cause mental health harms
By
–
It looks great. I assume it would be conflation but also linking how said racism could present itself in an anthropomorphic way aka Anthropic stuff today would be helpful. Tools pretending sentience that utilize racist datasets could cause unique mental health harms.
-
Mitigating Prompt Injection Through Data Logic Separation
By
–
The way to mitigate prompt injection is a clean separation between data and logic.
-
AI Safety Letter Sparks Critical Commentary and Debate
By
–
I realize the latest open letter from the "AI Safety" crowd is essentially self-parody, but at the same time, I couldn't resist … https://
docs.google.com/document/d/1Z_
Vz9rYUkajKbT2CF0SpFtfF8QspsHFKh9ozBFANseM/edit?usp=sharing
… (For the source reference, repair this URL: https: //openletter.svangel.com/ ) -
Uncanny Valley: AI Advancements and Psychological Anxieties
By
–
The Uncanny Valley: Advancements and Anxieties Of AI That Mimics Life
— Bernard Marr (@BernardMarr) 5 mars 2024
Exploring the psychological phenomenon of the #uncanny #valley in the age of #AI, we delve into the advancements and #anxieties surrounding #lifelike #robots and #digitalavatarshttps://t.co/JYnA3FR2NC pic.twitter.com/dJD1Y2CeeXThe Uncanny Valley: Advancements and Anxieties Of AI That Mimics Life Exploring the psychological phenomenon of the #uncanny #valley in the age of #AI, we delve into the advancements and #anxieties surrounding #lifelike #robots and #digitalavatars https://
bernardmarr.com/the-uncanny-va
lley-advancements-and-anxieties-of-ai-that-mimics-life/
… -
AI-Generated Fake News: Fabricated Quotes and Ethical Implications
By
–
I appreciate the opportunity to speak with an actual journalist (
@karanhm134
) about my experience finding a fabricated quote attributed to me in what turned out to be a fully fabricated (using the Gemini system) "news" article — and the broader implications of these practices: