Yes, LLMs are bad at: word games (due to tokenization), math (unless using tools), accurate citations & quotes, etc. Easy to find failures there The whole idea of the Jagged Frontier is that they are bad in ways you may not expect and good at other tasks that are hard for people
ETHICS
-
Meritocratic AI Systems Balancing Competence and Self-Actualization
By
–
Imagine you are looking through everyone's eyes simultaneously, but not everyone has the same skills and traits. You will want the system to be meritocratic (roles should be assigned according to competence), but it should maximize self actualization and integration for all.
-
System Incentives Over Incompetence: Why Fixing Systems Is Hard
By
–
Never explain by incompetence and malice what can be explained by people following their actual incentives. Fixing incentives is hard, because the system has no outside
-
Just Society Design: Universal Switching Test Ethics
By
–
Imagine you waking up in my body tomorrow, but with my memories, traits and skills, so nothing changes. I believe in a legal system and general societal rules that are designed in such a way that every individual would be ok switching places with every other individual.
-
US Government Control Over Global Tech Platforms: A Power Issue
By
–
If this were about privacy & security, then they'd be targeting Facebook, Instagram, etc as well. But they're not. This's about control & I'm not ok with ANY government exercising such control, particularly (as the US here) over the 'rest of world' via US-homed Big Tech platforms
-
Bill threatens foreign tech bans beyond TikTok platform
By
–
I'm surprised to see so many fans of this bill in the old mentions! So let me spell it out: This isn't just about TikTok. The bill gives USG the power to ban any old foreign-owned media/tech they designate as adversarial (while being totally OK w/FB etc serving 'the world')…
-
Privacy Enforcement Hypocrisy: Why Big Tech Remains Unregulated
By
–
if this were about privacy facebook, instagram, etc would also be banned.
-
Monica Lam: Conversational AI and Privacy Protection Research
By
–
CS prof. @MonicaSLam researches conversational virtual assistants with an emphasis on privacy protection. She's also an expert in compilers for high-performance machines. Her team created the first widely adopted research compiler, SUIF. 7/n https://
suif.stanford.edu/~lam/ -
Selective Privacy Regulation: Big Tech Accountability Double Standards
By
–
If this were about privacy and security then we'd be banning facebook, instagram, etc. as well.
-
AI Development Raises Serious Safety and Ethics Concerns
By
–
This is both terrifying and deeply, deeply stupid.