Having the code written by an AI known to its users to be AI is not the concern here; this isn't a labor thing or a mystical contamination, it's "Are these words I'm reading right now written by a human."
@esyudkowsky
-
AI content warnings ineffective due to poor implementation
By
–
I guess I should explicitly clarify–because of Earth's habit of picking exactly the worst possible way to do anything–that this is not AT ALL helped by having every website display a banner that says "May contain AI content somewhere."
-
Credibility concerns and regulatory capture risks in AI governance
By
–
Lacks credibility given their past actions. Politicians may assume it's a regulatory capture move. Seems more helpful to get Geoffrey Hinton in the room instead.
-
OpenAI Cannot Stop the AI Avalanche They Started
By
–
You're talking to the wrong person. OpenAI has zero ability to stop the avalanche they started. That's now a matter for treaties between major powers.
-
Human Flourishing as Optimization Problem and Gradient Descent Limitations
By
–
Nope, it's about human flourishing being an optimum of almost-zero utility functions and gradient descent looking like a terrible way to control inner preferences.
-
Detecting Additional Forces Beyond Token Prediction in Fine-tuned LLMs
By
–
Many LLMs have already been RLHFed and finetuned into activities other than "predict the next token a human would write". This being the case, how would you tell if the output was being driven by some extra force above that and all the finetuning?
-
The Inscrutability of LLM Low-Level Floats and Knowledge Gaps
By
–
You can! But in the case of LLMs, we don't. We speak of the inscrutability of the low-level floats because we have no higher-level knowledge to speak of instead.
-
Eliezer Yudkowsky denies wanting to control ASI
By
–
Who told you the lie that I wanted to centralize ASI or control it? I want it to not exist. I cannot control it. It kills everyone regardless of who thinks they control it.
-
AI industry downplays existential risks despite doom rhetoric
By
–
I cannot think of any case in history where "our product will kill everyone you love" was a good marketing tactic, and when Nvidia, the major winner so far at $1T cap, testified before the US Senate, they pooh-poohed talk of AGI (not just doom). You have been sold quite the bill