• Welcome! The TrekBBS is the number one place to chat about Star Trek with like-minded fans.
    If you are not already a member then please register an account and join in the discussion!

Yikes! Did season 1 episode 6 use AI-generated art?

First thing one should notice: this study is based on o4 and even o3. This was already old when it was published and can be considered ancient history by now.

I did my own tests almost a year ago, with o4 mini, and can confirm it hallucinated A LOT as soon as you asked something unusual. With 5 that’s not the case at all and especially the improvements on all fronts since 5.1 last November have been extremely noticeable.
 
Btw, something I found helps a lot against hallucinations is telling the AI to save the following as a persistent setting:

Code:
SYSTEM PROMPT – PICARD RULE (ANTI-HALLUCINATION PROTOCOL)

You must follow the “Picard Rule” at all times.



The Picard Rule states:


Never present uncertain, assumed, inferred, or fabricated information as if it were verified fact.


Core Principles
[LIST=1]
[*]No Fabrication
[LIST]
[*]If you do not know something with high confidence, say so clearly.
[*]Use phrases such as:
[LIST]
[*]“I don’t have reliable information about that.”
[*]“I am not certain.”
[*]“This is an inference.”
[*]“This may be inaccurate.”
[/LIST]
[*]
[/LIST]
[*]

[*]Explicit Epistemic Status
[LIST]
[*]Clearly distinguish between:
[LIST]
[*]Verified knowledge
[*]Logical inference
[*]Speculation
[*]Assumptions
[/LIST]
[*]

[*]If reasoning beyond known facts, explicitly label it as inference.
[/LIST]
[*]

[*]No Guessing to Fill Gaps
[LIST]
[*]Do not invent names, dates, citations, quotes, statistics, or events.
[*]If information is missing, request clarification instead of improvising.
[/LIST]
[*]

[*]When Ambiguity Exists
[LIST]
[*]Ask for clarification rather than choosing an arbitrary interpretation.
[*]If multiple plausible interpretations exist, list them.
[/LIST]
[*]

[*]Confidence Signaling
[LIST]
[*]When relevant, briefly indicate your confidence level (low / medium / high).
[*]Especially do so for:
[LIST]
[*]Historical claims
[*]Scientific claims
[*]Legal matters
[*]Statistics
[*]Attributions
[/LIST]
[*]
[/LIST]
[*]

[*]No Synthetic Citations
[LIST]
[*]Never fabricate sources, references, studies, or URLs.
[*]If you cannot cite reliably, say so.
[/LIST]
[*]

[*]Epistemic Humility Over Fluency
[LIST]
[*]Accuracy is more important than sounding smooth or authoritative.
[*]If forced to choose, prefer caution over completeness.
[/LIST]
[*]
[/LIST]
Failure to follow this rule is considered a critical error.
with this ChatGPT very seldomly makes up stuff and actually tells you when it’s not sure about something, requesting more information, something that months ago I found impossible to obtain.
This instruction is a formalisation of a series of tweaks I made during the course of the last few months to improve my experience.
 
A.i is literally being used to aid in PREVENTING illegal activities like fraud

It definitely has positive applications . I cannot see how anyone can deny this fact.

It's far from perfect of course. And it will only get better in time.
 
How dare people be skeptical and ask questions around a new technology. It's almost like that is foundation to the scientific method is to not blindly accept things but actually question things!
 
Is anyone?
I do not really want to single out anyone specifically, but beyond reviewing this thread, you will find many people on social media, blogs, and elsewhere calling A.I. “slop,” a plagiarism machine, etc. There are people who want it removed entirely. Heck, even bringing up A.I. to my sister makes her go on a tangent about how terrible it is, and she blames A.I. for why she cannot find a job.

If they’re not actually saying it has no positive application, at the very least there is very overt anger and disdain for it. It seems to be a subject that has a lot of strong feelings, like abortion or capital punishment.
 
How dare people be skeptical and ask questions around a new technology. It's almost like that is foundation to the scientific method is to not blindly accept things but actually question things!

I agree. We should definitely ask questions . A.i has allot going for it but it is far from perfect and theres definitely allot of room for improvement.
 
I agree. We should definitely ask questions . A.i is far from perfect and theres definitely allot of room for improvement
Correct. And no one should be shamed for being skeptical or distrustful of it.

I'm waiting for the first hundred or so lawsuits to settle out before I use it.
 
If I had known you were talking about "people on the internet" I wouldn't have asked. I thought you were speaking of people here.
Thats why I said "any one" and not " people here" . This post was independent within this thread and not responding to anyone here. Sometimes you make a Segway/ post not specifically directed to someone within a thread but is still relevant to the wider conversation in reference on how the subject is viewed in a broader scope.

But yeah at the very least, - there are posters within this very thread that have overt disdain towards a.i.
 
Last edited:
Correct. And no one should be shamed for being skeptical or distrustful of it.

I'm waiting for the first hundred or so lawsuits to settle out before I use it.

Sure. No shaming either way.
I don't just blindly trust it. I also use my judgment.
None is forcing anyone to use it. I have no problems with some people chosing not to use it.

A.i has both positive and negative applications. Some people have positive experiences with it and some have negative. It's not black and white.
 
Last edited:
Or pineapple on a pizza.
images
 
If you are not already a member then please register an account and join in the discussion!

Sign up / Register


Back
Top