AI usage in responses

[The Bread Guy]

I do hear of some people using AI to help write assessments, promotional packages, etc. It’ll be a cold day in hell before I use it for anything impacting the well being of my people or my own career advancement.

Bang on - no matter who/what analyzes/synthesizes for you, you know you’re where the buck stops.

I don’t have enough knowledge for how well use it in business intelligence applications but I’m sure it has lots of potential there.

I’m still learning, too, but from an OSINT perspective, here’s some decent stuff from folks I trust talking about have a tool belt of AI platforms doing what each is best at. Yeah, they’re selling their seminars, too, but still nuggets to be mined here.

Building Your First AI Stack: Start Small, Think Strategically

Begin with one tool, one task, and one practical need.

substack.com

🤖 AI Tools for OSINT: A Practical Starting Point for Investigators

An introduction to where AI tools fit in investigative workflows

toddingtonint.substack.com

 

[FSTO]

I use AI to help clean up my initial drafts.

But I do worry that I'm leaning more into the AI and writing Human engineered slop in the first place.

 

[kratz]

I have crossed paths with start up businesses selling AI an automated decision making tool for emergency and disaster events. I could use the current provincial state of emergency in BC as an example of where these people want to see their AI deployed. AI is a tool, but I object when these programmers do not know or understand the field which they are attempting to sell their products to. I ask the rhetorical question; Who will shoulder the responsibility for the inevitable failed decisions by AI in these situations?

 

[FSTO]

I have crossed paths with start up businesses selling AI an automated decision making tool for emergency and disaster events. I could use the current provincial state of emergency in BC as an example of where these people want to see their AI deployed. AI is a tool, but I object when these programmers do not know or understand the field which they are attempting to sell their products to. I ask the rhetorical question; Who will shoulder the responsibility for the inevitable failed decisions by AI in these situations?

 

[The Bread Guy]

I have crossed paths with start up businesses selling AI an automated decision making tool for emergency and disaster events. I could use the current provincial state of emergency in BC as an example of where these people want to see their AI deployed. AI is a tool, but I object when these programmers do not know or understand the field which they are attempting to sell their products to. I ask the rhetorical question; Who will shoulder the responsibility for the inevitable failed decisions by AI in these situations?

We already know the AI corporate answer if all the legacy systems say, “hey, we can fuck up” as a standard disclaimer. As others upthread & elsewhere have said, you decide, you sign, you own.

 

[mariomike]

I stick with the library , partly because I am old-fashioned, but also because professional librarians vet the fiction from non-fiction.

So, I don't have to.

I find that preferable to watching AI Slop / Rage Bait, and trying to figure out what is fact, and what is fiction.

Readers with an interest in Artificial intelligence may, or may not , find this book of interest.

Unmasking AI — Toronto Public Library

Unmasking AI — Buolamwini, Joy, — "Dr. Joy Buolamwini is the self-described "Poet of Code" who has had a lifelong passion for computer science, engineering, and art-disciplines that, she felt, pushed the boundaries of reality. After tinkering with robotics as a high school student in Tennessee...

tpl.bibliocommons.com

 

[The Bread Guy]

Don't be a "meat proxy"

Are You a 'Meat Proxy'? The Rise of Copy-Pasting AI Responses Without Thinking - CNET

"AI slop" now has a friend. "Meat proxy" is a new way to describe too much reliance on AI content.

www.cnet.com

Don't be a meat proxy

 

[mariomike]

Don't be a "meat proxy"

Interesting definition,

https://twitter.com/x/status/2092089744050061717

 

[Altair]

Interesting definition,

https://twitter.com/x/status/2092089744050061717

can we make that a policy here?

 

[mariomike]

I read about a 30% AI rule in education .

That students limit artificial intelligence to 30% .

The remainig 70% their own original thinking and research.

 

[Altair]

I read about a 30% AI rule in education .

That students limit artificial intelligence to 30% .

The remainig 70% their own original thinking and research.

Can we make that a policy here?

 

[markppcli]

It offers an opinion. I have known encyclopedias and dictionaries to revise their entries. Even the Bible for that matter.

The problem is that the AI spam wall post is presented not as opinion but as fact. I think at a minimum AI posts should always include the prompt used. As others have eluded to ChatGPT, and other AI, save your data and history to "improve interactions." Its given you an editorializing view based on your previous questions and any linked social media - so for example Grok will take into account your X history. So yes opinion, and it can gather raw data, but the prompt and history will alter what comes out.

 

[The Bread Guy]

... the prompt and history will alter what comes out.

That right there, just like with search engines.

 

[Kirkhill]

The problem is that the AI spam wall post is presented not as opinion but as fact. I think at a minimum AI posts should always include the prompt used. As others have eluded to ChatGPT, and other AI, save your data and history to "improve interactions." Its given you an editorializing view based on your previous questions and any linked social media - so for example Grok will take into account your X history. So yes opinion, and it can gather raw data, but the prompt and history will alter what comes out.

That right there, just like with search engines.

And that, in a nutshell, defines the AI answer.

It is like working with a particularly obtuse librarian with pre-conceived notions. It answers exactly the question you ask based on what it thinks it knows, or has been instructed to release.

Sometimes you have to rephrase, reset and reask multiple times to find information you know exists and have that information incorporated into the answer, the answer you want. And obviously the answer I want is the right answer.

In my opinion.

....

Ancient history.

When I was being taught composition in primary school there were three words I could not use: "I", "Got", and "Believe".

Everything was written in the third person. No "I".
"Got" was a place holder that covered too many sins and showed a lack of imagination and vocabulary.
And everything written was assumed to be nothing more than the opinion of the author therefore "I believe" was redundant.

Facts were in short supply.
You supported arguments with corroborating opinions from other people.

 

[brihard]

The problem is that the AI spam wall post is presented not as opinion but as fact. I think at a minimum AI posts should always include the prompt used. As others have eluded to ChatGPT, and other AI, save your data and history to "improve interactions." Its given you an editorializing view based on your previous questions and any linked social media - so for example Grok will take into account your X history. So yes opinion, and it can gather raw data, but the prompt and history will alter what comes out.

I like the prompt disclosure aspect.

The AI and Geopolitics course I just finished encouraged use of AI for assignments, but a key part was disclosing our prompt and response history. It was interesting learning from trial and error how much work went into generating good prompts that would return useful product.

If in time I should find myself using AI for any of my posts here, my prompt(s) and the AI tag will accompany.

 

[The Bread Guy]

The AI and Geopolitics course I just finished encouraged use of AI for assignments, but a key part was disclosing our prompt and response history. It was interesting learning from trial and error how much work went into generating good prompts that would return useful product.

Plugging the same prompt into different systems can also show interesting results, too.

... I think at a minimum AI posts should always include the prompt used ...

Very good best practice, too.

 

[mariomike]

Can we make that a policy here?

Opinions over these three pages regarding posting AI vary. As far as I know , there is no Policy .