• Please take a moment and update your account profile. If you have an updated account profile with basic information on why you are on Air Warriors it will help other people respond to your posts. How do you update your profile you ask?

    Go here:

    Edit Account Details and Profile

Are you using AI - and how?

Hair Warrior

Well-Known Member
Contributor
NIPR has ChatGPT now. Test out your FITREP bullets and 1650s to your heart’s content.

 

Brett327

Well-Known Member
None
Super Moderator
Contributor
NIPR has ChatGPT now. Test out your FITREP bullets and 1650s to your heart’s content.

DoN still has a policy prohibiting GPT on official work, AFAIK.
 

Spekkio

He bowls overhand.
DoN still has a policy prohibiting GPT on official work, AFAIK.
Yeah, and I also can't drive faster than 65 mph on a highway (75 mph west of Nebraska-ish). But today I drove 95 mph along with several other drivers near me, and the world seems to turn just fine. Even passed a few police officers along the way who apparently weren't interested in pulling people over for speeding.

Point being: I'm interested in how you think the DoN is going to enforce a zero tolerance policy for GPT use in any capacity for official business, and why this policy is beneficial for the force. As a major commander on this forum, you're awfully good at throwing spears at others, yet rarely make a cogent argument for any particular policy position. You're basically Jon Stewart with significantly less entertainment value.
 
Last edited:

Spekkio

He bowls overhand.
NIPR has ChatGPT now. Test out your FITREP bullets and 1650s to your heart’s content.

Ok.... first an anecdote...

So my kids downloaded an AI image generator, and for shits and giggles I wanted to create an image of myself and my wife doing something. So the first pass returned some images of blonde people, and we are not... so I input "brown haired..." Well, it turns out that the AI returns images with only men with beards, which I - as a servicemember - don't have. So I put the constraint 'no beard' but AI still returns men with beards. Also, at this point I noticed that the men all had the same hairstyle.

So I put in middle aged (yuck, I'm over 40) and AI returned people with 100% gray hair that looked 60+.

I put in our ethniticies and rather than melding our appearance and facial structure to that, they put us in ridiculous outfits representing those cultures.

Where I'm going with this is that it's insightful how AI has been "trained" what an attractive man and woman looks like and any deviation is "wrong." And this phenomenon can be applied to anything AI does, including FITREPs.

There are a lot of shitty fitrep bullets floating around open source. Beyond getting a good hard and soft breakout, a good fitrep links an officer's contributions to advancing broader TYCOM, COCOM, or OPNAV priorities (which change over time), because that's how the selection boards derive their guidance. Even at the O3 / O4 level, you can write about how your participation in exercises contributed to warfighting initiatives (bonus if you can use the words 'joint' or 'combined'). No one gives a shit that you improved your division's PAR from 85% to 95%. In other words, you should focus on your 'up and out' contributions and not 'down and in' accomplishments. Given the amount of junk fitrep bullets floating around the JO gouge underground, chat GPT is unlikely to give you the quality you need to advance your career. But good luck with that.
 
Last edited:

Hair Warrior

Well-Known Member
Contributor
Appreciate the share! This memo does not prohibit the use of NIPRGPT. On the contrary, it appears NIPRGPT was designed specifically to address many of the same risks pointed out in this memo.

NIPRGPT is not Commercial AI. It is rated to IL4 CUI. It cannot be used for PII/PHI or for personal use. And if you log in with CAC it has certain caveats and user agreement provisions stating that a human must review and be accountable for use of any NIPRGPT output.
 
Last edited:

Brett327

Well-Known Member
None
Super Moderator
Contributor
I'm interested in how you think the DoN is going to enforce a zero tolerance policy for GPT
Same way they do every other directive... with mixed results. I'm not wagging my finger, more so reminding people that DoN doesn't yet have a coherent policy on how to use GPT-like capabilities. As a major commander on this forum, it isn't my job to come up with cogent arguments for how any Navy policy should be implemented. That's what your ISIC is for, and I'd love to be in the room when you pitch them your innovative "ignore all policies I don't like" paradigm of leadership. :D
 

Spekkio

He bowls overhand.
I, too, have airwarriors'd after a couple two three whiskeys. 😜
Well, the tl: dr is chat GPT is basically a crowd sourcing app. If the crowd thinks the earth is flat, AI will also spit that out as fact. And once it's trained by big data, you can't tell it any different.

Most fitreps are not EP fitreps (by instruction). So chat GPT is going to think a P writeup is "correct."
 
Last edited:

Spekkio

He bowls overhand.
Same way they do every other directive... with mixed results....
That's awfully generic.

How are you going to identify that someone used chat GPT to generate parts of their fitreps or any other official documents? "This reads like it was written by chat GPT" isn't going to be good enough. It's a virtually impossible rule to enforce.

That's what your ISIC is for, and I'd love to be in the room when you pitch them your innovative "ignore all policies I don't like" paradigm of leadership. :D
Did you not allow deployment beards when in command?
 
Last edited:

Fins Out

Well-Known Member
I think the current FITREP system already prevents ChatGPT and the like from having any influence. The used car salesman block of my FITREP could state the cold fusion reactor I designed can power the entire planet, cure cancer AND erectile disfunction. But, as an O-4, unless my high water hard breakout is a #1 EP, I'm probably not being selected for O-5 command.
 
Top