OdoHeDidNot (he/him)
banner
odohedidnot.bsky.social
OdoHeDidNot (he/him)
@odohedidnot.bsky.social
Sentient liquid god, top dog on the promenade.
probably be motivated by a desire to provide an accurate answer (or to at least respond with a specific amount of accuracy). If you couldn't provide an accurate answer, you might even refuse to answer at all.

These response strategies are contingent on types of awareness unavailable to LLMs.
August 7, 2025 at 9:02 PM
you a question, you probably aren't going to choose your words based on how likely those words would show up in a response to that question given all of the responses to all the questions available on the internet.

Instead, you'd consider social context, relative expertise, mood, etc. You'd also
August 7, 2025 at 8:58 PM
No actual particle physicist would likely be fooled for long.

These models don't "think" about a prompt, they process a prompt and spit out a prediction of what a human might say in response to that prompt. This is not what happens when a human brain is prompted for a response.

When someone asks
August 7, 2025 at 8:53 PM
If you spent enough time eavesdropping on conversations between particle physicists, you could probably convince a layperson you were a particle physicist purely by mimicking the words you overheard. LLMs extend this dynamic by "eavesdropping" on as much human generated text as possible.
August 7, 2025 at 6:18 PM
an LLM alone will never produce anything truly "intelligent" in the way that biological organisms are intelligent.

These models are designed to produce sequences of words that mimic sequences of words produced by cognitive processes, but the models don't actually replicate cognitive processes.
August 7, 2025 at 6:11 PM
Does this case demonstrate an exception to the parol evidence rule or was it a judicial misstep that's since been rectified? The textbook wants to know what this 1.5L thinks.

I have words for whoever decided that the Socratic Method was an appropriate approach to textbook authorship.
March 30, 2025 at 7:47 PM
I believe that. This post was inspired by my Property casebook.
January 19, 2025 at 10:34 PM
📌
November 25, 2024 at 3:49 AM
World history is littered with less than hypothetical examples of dehumanizing speech building support for and facilitating genocide. It could easily be argued that this type of speech is getting louder right now in the US. There is nothing dishonest about this framing.
January 10, 2024 at 10:37 PM
On the other hand, when it comes to defending the right of a group of people to advocate for your murder, do you cease their defense before your neck is in the gallows or after?
January 10, 2024 at 10:24 PM
At the gate of the flight you want to be on has worked for me.
December 29, 2023 at 3:49 PM
He 100% tried to buy Wikipedia and they told him to pound sand and now he's mad about it.
October 23, 2023 at 12:11 AM