Good call, it was actually the “Button Shapes” setting
June 4, 2025 at 7:01 AM
Good call, it was actually the “Button Shapes” setting
> If I wanted to discuss NPD, for example, I would use the DSM-5, not either of those
I don’t think those are mutually exclusive, the second reference in the Wikipedia article for NPD is the DSM-5
I guess: if I want to dispute the claim “LLMs are not AI”, where do you think should I source from?
I don’t think those are mutually exclusive, the second reference in the Wikipedia article for NPD is the DSM-5
I guess: if I want to dispute the claim “LLMs are not AI”, where do you think should I source from?
May 20, 2025 at 3:58 PM
> If I wanted to discuss NPD, for example, I would use the DSM-5, not either of those
I don’t think those are mutually exclusive, the second reference in the Wikipedia article for NPD is the DSM-5
I guess: if I want to dispute the claim “LLMs are not AI”, where do you think should I source from?
I don’t think those are mutually exclusive, the second reference in the Wikipedia article for NPD is the DSM-5
I guess: if I want to dispute the claim “LLMs are not AI”, where do you think should I source from?
You assume it’s not my field as well?
When we’re talking basic definitions I think Wikipedia is a fine source. I could’ve used the dictionary too, would that make you happier?
Here’s the OED for “artificial intelligence”
But I’m sure you’re going to say “don’t use the dictionary for definitions”
When we’re talking basic definitions I think Wikipedia is a fine source. I could’ve used the dictionary too, would that make you happier?
Here’s the OED for “artificial intelligence”
But I’m sure you’re going to say “don’t use the dictionary for definitions”
May 20, 2025 at 3:36 PM
You assume it’s not my field as well?
When we’re talking basic definitions I think Wikipedia is a fine source. I could’ve used the dictionary too, would that make you happier?
Here’s the OED for “artificial intelligence”
But I’m sure you’re going to say “don’t use the dictionary for definitions”
When we’re talking basic definitions I think Wikipedia is a fine source. I could’ve used the dictionary too, would that make you happier?
Here’s the OED for “artificial intelligence”
But I’m sure you’re going to say “don’t use the dictionary for definitions”
I am once again begging you to read Wikipedia
May 20, 2025 at 4:21 AM
I am once again begging you to read Wikipedia
That’s not what any of those words mean. Really, you can just start by reading Wikipedia!
en.m.wikipedia.org/wiki/Artific...
en.m.wikipedia.org/wiki/Artific...
May 20, 2025 at 4:16 AM
That’s not what any of those words mean. Really, you can just start by reading Wikipedia!
en.m.wikipedia.org/wiki/Artific...
en.m.wikipedia.org/wiki/Artific...
By definition LLMs are AI. It’s trivial proven by going to the Wikipedia pages for “Large language model” and “machine learning”, which I’ve screenshotted below.
LLMs are machine learning models, and machine learning is a field within AI.
LLMs are machine learning models, and machine learning is a field within AI.
May 20, 2025 at 3:25 AM
By definition LLMs are AI. It’s trivial proven by going to the Wikipedia pages for “Large language model” and “machine learning”, which I’ve screenshotted below.
LLMs are machine learning models, and machine learning is a field within AI.
LLMs are machine learning models, and machine learning is a field within AI.
Maybe you should’ve read the Wikipedia links it gave you before trying to dunk on the answer. The “ideology” dimension is separate from the “autocracy” dimension, so you can definitely have a center right dictatorship.
May 19, 2025 at 8:15 PM
Maybe you should’ve read the Wikipedia links it gave you before trying to dunk on the answer. The “ideology” dimension is separate from the “autocracy” dimension, so you can definitely have a center right dictatorship.
Ok now we can discuss AGI
May 18, 2025 at 10:07 PM
Ok now we can discuss AGI
I can see that it did run a search for my MLB query, and it’s actually useful in debugging what went wrong. The query was probably too restrictive because it used its prior knowledge about the Mets and that surfaced articles from when the White Sox tied their record, missing that they broke it later
May 16, 2025 at 1:58 AM
I can see that it did run a search for my MLB query, and it’s actually useful in debugging what went wrong. The query was probably too restrictive because it used its prior knowledge about the Mets and that surfaced articles from when the White Sox tied their record, missing that they broke it later
The o3 model can perform web searches so it should be able to get it. Like I can ask it “Who is the pope?” and it tells me Leo XIV, not Francis.
May 16, 2025 at 1:58 AM
The o3 model can perform web searches so it should be able to get it. Like I can ask it “Who is the pope?” and it tells me Leo XIV, not Francis.
I’ll be honest, I tried to debunk this and it also failed for me. For my query it thought that the 2024 White Sox tied the 1962 Mets at 120 losses, but that’s wrong because the White Sox really had 121.
May 16, 2025 at 1:44 AM
I’ll be honest, I tried to debunk this and it also failed for me. For my query it thought that the 2024 White Sox tied the 1962 Mets at 120 losses, but that’s wrong because the White Sox really had 121.
When OpenAI releases models even the blog posts for the general public have eval numbers in them and I’ve never seen one that says 100%. They’re overhyping their models but I can’t see anywhere they’ve said they don’t make mistakes. Here’s one graph from the latest one
openai.com/index/introd...
openai.com/index/introd...
May 7, 2025 at 2:01 AM
When OpenAI releases models even the blog posts for the general public have eval numbers in them and I’ve never seen one that says 100%. They’re overhyping their models but I can’t see anywhere they’ve said they don’t make mistakes. Here’s one graph from the latest one
openai.com/index/introd...
openai.com/index/introd...
And the ChatGPT terms of service say this
May 7, 2025 at 1:55 AM
And the ChatGPT terms of service say this
Let me Wikipedia that for you. LLMs are obviously predictive. If you have a problem with how they’re being “sold”, you should be clearer about that. There are plenty of use cases for them if you can look beyond the marketing hype (which does exist!)
May 6, 2025 at 9:54 PM
Let me Wikipedia that for you. LLMs are obviously predictive. If you have a problem with how they’re being “sold”, you should be clearer about that. There are plenty of use cases for them if you can look beyond the marketing hype (which does exist!)
How long has it been since you tried it? Both gpt-4o and o3 do fine with that prompt today
May 5, 2025 at 10:50 PM
How long has it been since you tried it? Both gpt-4o and o3 do fine with that prompt today
You might be using an old version, this is what I get in the app today
May 5, 2025 at 9:04 PM
You might be using an old version, this is what I get in the app today
This just isn’t true anymore
May 2, 2025 at 5:06 PM
This just isn’t true anymore
In 2009, PageRank was the worst-performing method that the Microsoft team tried in the TREC competition, well below BM25, a standard term-based retrieval method
trec.nist.gov/pubs/trec18/...
trec.nist.gov/pubs/trec18/...
May 2, 2025 at 4:05 AM
In 2009, PageRank was the worst-performing method that the Microsoft team tried in the TREC competition, well below BM25, a standard term-based retrieval method
trec.nist.gov/pubs/trec18/...
trec.nist.gov/pubs/trec18/...
Ok, to be more exact I should’ve said the model is “good at using the search tool”, if that makes a difference.
From what I can see, ChatGPT meets the Wikipedia definition of search engine. I send it a question and it sends me back links with information that answers the question.
From what I can see, ChatGPT meets the Wikipedia definition of search engine. I send it a question and it sends me back links with information that answers the question.
May 1, 2025 at 10:27 PM
Ok, to be more exact I should’ve said the model is “good at using the search tool”, if that makes a difference.
From what I can see, ChatGPT meets the Wikipedia definition of search engine. I send it a question and it sends me back links with information that answers the question.
From what I can see, ChatGPT meets the Wikipedia definition of search engine. I send it a question and it sends me back links with information that answers the question.
Got blocked for telling someone that GenAI is one kind of ML
April 30, 2025 at 4:15 AM
Got blocked for telling someone that GenAI is one kind of ML
Being wrong? If you’re trying to say “other kinds of ML are more useful than GenAI”, just say that instead. That’s a debate you can actually have because it’s an opinion and not just an outright falsehood
April 30, 2025 at 4:07 AM
Being wrong? If you’re trying to say “other kinds of ML are more useful than GenAI”, just say that instead. That’s a debate you can actually have because it’s an opinion and not just an outright falsehood
Anyway, I think your premise is incorrect. I asked ChatGPT which threads it meant and since o3 has real-time search I think it might’ve been referring to this thread, see one of the links in the screenshot
April 29, 2025 at 6:28 PM
Anyway, I think your premise is incorrect. I asked ChatGPT which threads it meant and since o3 has real-time search I think it might’ve been referring to this thread, see one of the links in the screenshot