Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
AI makes modern art pieces like the guy who simply knocked over some sand bucket…
ytc_Ugw9T-h_d…
G
Gemini gave the same answer!
Muhammad.
CoPilot is afraid and said I can't help …
ytc_UgxJNB_69…
G
And these "autonomous" operation modes are beginning to show up on cargo transpo…
ytc_UgxyjOF89…
G
It could be fixed. This is my ideas. We need a law. That no company owner can bu…
ytc_UgzNIUb_6…
G
ChatGPT is an actor with a library card. Can an actor with the world's knowledg…
ytc_UgzeI58wE…
G
It is nonsense that we do not know how AI works. The system is designed by peopl…
ytc_UgyU-Dj6-…
G
Dude all of your takes are hyperbolic and not in line with reality. Can you stop…
ytc_UgxeW8EZ7…
G
10:44 Seeing this made me feel like puking.. as an afab (assigned female at birt…
ytc_UgyzUGyO9…
Comment
I like to give chatgpt a portion of code, ask it to describe how it works currently. Then I tell it what I want to add on, or what I'd like to modify, and what the goal is I'm trying to accomplish. I get pretty detailed, sometimes 2-3 paragraphs with examples. Then I ask gpt if my solution solves the problem, if it presents any new potential problems, and if there is a better/more efficient/secure etc way to achieve the goal.
Another great way to use gpt for development is brainstorming. If you need ideas, or the project feels like it's going stale, you can give chatgpt the code, and then tell it to interview you like a senior developer. I'll give you an example prompt that I used recently.
This was for a game I'm making in my free time.
First I uploaded the code in a project. Then I asked gpt to "describe the current gameplay" inside that project.
Then I asked it to generate a prompt I could use to replicate the game.
Then I asked it to describe how a specific item in the game works, a really simple drop that boosts player hp.
That was the setup to get gpt "aligned" with what I want and where I'm going. Do this and you'll get much better results.
It described the way the item worked correctly, and I affirmed it's description. Then said:
(prompt 1)
"now we need to move on to weapons. this area needs a lot of development. the current weapon system needs to be entirely rewritten, as I don't like it.
here's what I'm envisioning.
the player has 4 slots they can use to equip different weapons. Each weapon has a "Type" that corresponds to the "type" of slot. For example,
-Melee
-Ranged
-Homing
-Special
A melee weapon can only be equipped in a melee slot, a ranged weapon in ranged, etc.
do you follow so far? don't jump ahead or make changes, just confirm whether or not you understand. if you have questions, ask"
Now I came in with the prompt. This prompt has two parts. The first part establishes the intent, the second part defines the process."
At this point, I was just spitballing ideas. The weapon system existed in theory, it wasn't implemented yet. As instructed, gpt confirmed it "followed" by restating my goal in simpler terms, and asking 7 questions about my request since the weapon system I was referring to didn't actually exist.
This is where the second half of the prompt comes in.
(prompt 2)
"lets address those questions one by one, interview style. I want you to ask me each of those one at a time, in whatever order you think is most efficient. At the end of the "interview," i want you to review my answers and reassess your understanding and ask follow up questions if needed. ready when you are"
It will do just that. Gpt will ask you each of it's original questions one at a time. These questions can change though, as gpt reevaluates the first prompt every time you answer a question.
Depending on the code base and complexity of your prompt setup, this chat session will keep going. The session I described here started with 5 questions and is still ongoing at question 47.
Granted I was using this as a brainstorming session, so I answered a lot of gpts questions with more questions, which definitely accounted for most of the replies.
But for me, my favorite way to use gpt is to have it ask me questions, and force me to really think through my answer.
youtube
AI Jobs
2025-08-26T15:5…
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | unclear |
| Policy | none |
| Emotion | approval |
| Coded at | 2026-04-27T06:24:59.937377 |
Raw LLM Response
[
{"id":"ytc_Ugzq-87VHQ4fbxm9FUl4AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_Ugyh8-9zauu1rls-BIN4AaABAg","responsibility":"ai_itself","reasoning":"deontological","policy":"none","emotion":"outrage"},
{"id":"ytc_UgxNNXe3JA_lZARa41J4AaABAg","responsibility":"ai_itself","reasoning":"consequentialist","policy":"none","emotion":"fear"},
{"id":"ytc_UgxoOcDAQztVOdy1EVJ4AaABAg","responsibility":"ai_itself","reasoning":"deontological","policy":"none","emotion":"mixed"},
{"id":"ytc_UgwIKlSd-iOpwYPoaNx4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugwzb8Hrt49IvghE_SR4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"mixed"},
{"id":"ytc_UgyOHm6EzaA_mB88jS14AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"approval"},
{"id":"ytc_UgyPG8Oym4uTduByWs94AaABAg","responsibility":"none","reasoning":"unclear","policy":"none","emotion":"indifference"},
{"id":"ytc_Ugwh4Qrw9gr8xR3kQJt4AaABAg","responsibility":"user","reasoning":"virtue","policy":"none","emotion":"mixed"},
{"id":"ytc_UgygpLDi6yqo008d2Dh4AaABAg","responsibility":"developer","reasoning":"consequentialist","policy":"none","emotion":"outrage"}
]