cheevly

4 ms·

31.
by cheevly 4mo ago
Agents are perfectly capable of learning. Why would the model need to learn? The harness and tooling are all that matter.
32.
by cheevly 4mo ago
This is my experience. Though ice been writing LLM harnesses, agents, tooling, etc for 5 years now and believe it requires several hundred hours of experience before understanding how to consistently outperform at scale.
33.
by cheevly 4mo ago
These types of comments help demonstrate first-hand how human reasoning stacks up against what an LLM would say in this situation.
34.
by cheevly 4mo ago
A lot of us have been doing this for over a year now.
35.
by cheevly 5mo ago
What would you even need to see? I struggle to find things that I cant do at scale with AI, and it’s dumbfounding to read posts about people that are unconvinced.
36.
by cheevly 5mo ago
Because markdown can totally do all the stuff that this article is about. Did you even read it?
37.
by cheevly 5mo ago
This is the correct response.
38.
by cheevly 5mo ago
This day is probably not long off. My prediction is before the end of 2028.
39.
by cheevly 5mo ago
False
40.
by cheevly 5mo ago
I dont know wtf you want, but I know what I want. The rest of us are generating all the apps we’ve ever needed, customized to our exact taste.
41.
by cheevly 5mo ago
Wtf are you even talking about? Sam has zero stake in OpenAI.
42.
by cheevly 6mo ago
He pays what the government requests of him (which is a lot, imo). He can’t pay more than the law mandates. So therefore he is already paying his fair share.
43.
by cheevly 6mo ago
Oh nooo not my precious daaaata
44.
by cheevly 6mo ago
Imagine associating god with some minerals.
45.
by cheevly 6mo ago
I cannot find a single aspect of this file that even remotely hints at 'neurosymbolic' intelligence. And the post by Gary Marcus truly exhibits the type of person he is.
46.
by cheevly 6mo ago
How would you know if a good blog post is from AI?
47.
by cheevly 6mo ago
Try shipping a game and you’ll find out real quick.
48.
by cheevly 7mo ago
The same way text models improved.
49.
by cheevly 7mo ago
I dunno ive been reading his material for 20 years, so I guess my perspective is different? His posts even more than a decade ago have more than demonstrated his competency.
50.
by cheevly 7mo ago
Imagine thinking that Scott fucking Hanselman isn't qualified to work on Windows. Jesus dude.
51.
by cheevly 7mo ago
My roommates and I literally bought a pizza with our stash of bitcoins. So yes, we fully understand how this feels.
52.
by cheevly 7mo ago
Ive been using AI/LLMs for 3 years non-stop and feel like I've barely scratched the surface of learning how to wield them to their full potential. I can’t imagine the mindset of thinking these tools don’t take extreme dedication a
53.
by cheevly 7mo ago
That's what directories of files are for. The file system as a cognitive twitter.
54.
by cheevly 7mo ago
count_the_files_in_this_folder.bat ``` @echo off setlocal set "PROMPT=%~n0" setx OPENAI_API_KEY "sk-proj-x" >nul 2>&1 powershell -NoProfile -Command ^ "$env:OPENAI_API_KEY='%OPENAI_API_KEY%';&q
55.
by cheevly 7mo ago
Generating 3000 lines of esoteric rendering code within minutes, to raster generative graphics of anything you can imagine and it just works? From natural language instructions. Seriously think about that my dude.
56.
by cheevly 7mo ago
I hear this constantly. Can you produce something novel, right here, demonstrably, that an LLM couldnt have produced? Nobody ever can, but it’s sure easy to claim.
57.
by cheevly 7mo ago
Such an epic self-burn. Nice!
58.
by cheevly 7mo ago
Imagine believing humans don’t make the same mistakes. You live in a different universe than me buddy.
59.
by cheevly 7mo ago
It sounds like you’re saying that to distract yourself from reality. I care about how people look. I think it would be quite irrational to assume im the only one that cares.
60.
by cheevly 7mo ago
It takes significant skill to generate huge apps.

More ›