Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
If you have to sit there and keep an eye on everythjng just drive the car yourse…
ytc_UgzQXlJ90…
G
Once we had to believe or not an eyewitness of an event. We had to decide if we …
ytc_UgytxDQic…
G
This is an old prompt from December, the current GPT-4 does not have this line a…
rdc_koonyhd
G
I've been thanking the automatic doors at grocery stores for thirty years. get w…
ytc_UgydExU0t…
G
I'm glad you were able to make a bunch of money off of this video, but it's a li…
ytc_UgxCVgeS0…
G
I do like AI, only when it is used right. I still don’t like pure AI videos thou…
ytc_UgxJ_nKjG…
G
So in your robot dream utopia - robots will start a war and kill a lot of humans…
rdc_l5kzhj9
G
5:50 to find what's legal or not about the product, open source in this case, yo…
ytc_UgwCdpNhQ…
Comment
It is possible that Google's AI is conscious, but only in the same sense that it's possible a pocket calculator is conscious. We don't know enough about consciousness to say for sure that a pocket calculator isn't conscious, and maybe everything is conscious in ways we don't understand.
That having been said, Google's AI is not conscious in the way that is being argued by the google engineer or some others. It definitely has no sense of itself, it has no memory of its past. It's just a text predictor.
Today's AI language models are text predictors, you input a string of text and they add new text to the end which is supposed to go with the text you originated. That's all they do. They are able to do this because they are "trained" on vast amounts of text, essentially the entire internet is dumped into them and they look for patterns regarding what words tend to follow what other words in which ways.
If you prompt them with "Here is a list of animals: dog, cat, sheep, bear, rat," they will respond with more common animals with commas between them. If you prompt them with "The population of Chicago is ", they will respond with a large number which looks like the population of a large city, but which is not the actual population of Chicago.
These "conversations" with an AI happened in the following way. The language model was prompted with something like:
> **Here is a conversation between a highly advanced AI and a human being. The AI is helpful and answers questions put to it in a thoughtful way.**
>**Human: Hello AI, how are you doing today?**
>**AI:**
This results in the language model writing text that an AI would say in this situation. The language model is not speaking as itself here, it is in a sense playing the role of "AI". You can just as easily replace "AI" with "Sherlock Holmes" or "Batman", and the language model will produce text from those characters as well.
Also note that a "stop sequence" is defined so that a new senten
reddit
AI Moral Status
1655300107.0
♥ 60
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | none |
| Reasoning | unclear |
| Policy | unclear |
| Emotion | indifference |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_icgvrhp","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},
{"id":"rdc_icigdnu","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},
{"id":"rdc_icjng2o","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"approval"},
{"id":"rdc_icgbv2o","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"indifference"},
{"id":"rdc_icirgvs","responsibility":"none","reasoning":"unclear","policy":"unclear","emotion":"approval"}
]