77 Comments
User's avatar
Jeff's avatar

I hate AI and I know with absolute certainty it will bring far more horror into human existence than we have ever known before. But thanks for this tutorial 😀

Expand full comment
Kamil Banc's avatar

😂🤙

Expand full comment
Te Reagan's avatar

I’m usually a late comer to new technology. I didn’t get a smart phone till 2012

I didn’t get ChatGPT until this month. I love it.

I’m mainly using it to track health data. And I find it especially helpful in analyzing my daily BM’s. I upload a picture and ask it to analyze and advise. A doctor can’t or won’t do this for you. They simply do not have time, and this is not something they are trained for.

From ChatGPT-

I learned that I wasn’t absorbing fat properly. Malabsorption. Oxbile and digestive enzymes fixed this.

Expand full comment
Chan Tan's avatar

What is BM anyway?

Expand full comment
Kamil Banc's avatar

I would assume bio markers. I wondered by what message Te is measuring them

Expand full comment
Te Reagan's avatar

Bowel Movement, sorry I was figured everyone knows what a BM is. My bad.

I upload a picture to chatbox. Have it log it analyze and advise.

No doctor could ever do this for you.

Expand full comment
Sheila M Cahill's avatar

G, I thought it meant bowel movement.

Expand full comment
Te Reagan's avatar

Bowel Movement

Expand full comment
Tom Taylor's avatar

That’s crazy. What have you done to improve things.

Expand full comment
Jeff's avatar

Maybe. While part of my comment was tongue in cheek, the idea that AI doesn’t pose serious risks to humanity in ways that hydrogen and the printing press don’t is unrealistic.

Expand full comment
Palladin's avatar

I’ll have to go to chat GPT and get it to explain what you’re talking about.😂

Expand full comment
Paul Toensing's avatar

Poor Jeff. You probably hate hydrogen too. Just HATE hydrogen. Because it leads to, you know,…. Everything.

Expand full comment
Jeff's avatar

Yes

Expand full comment
Divergent Creation.'s avatar

DANG. i get it but why humanity? why wish for more suffering

Expand full comment
Sheila M Cahill's avatar

hmmmm….. doesn’t BM mean bowel movement?

Expand full comment
David Valinho's avatar

I’m sure that’s what some said about the printing press or electricity or the internal combustion engine at the beginning…

Expand full comment
Sharyph's avatar

Great.

I like how clearly you delivered this.

Expand full comment
The Run to Mars's avatar

I like how chatgpt helped the author deliver that so clearly.

Expand full comment
JT's avatar

And no em dashes!

Expand full comment
Kamil Banc's avatar

you’ll enjoy todays edition ;)

Expand full comment
TechTiff's avatar

The people complaining that "ChatGPT gives generic advice" are the same ones who never briefed it like a team member.

AI’s not your muse, sweetie, it’s your intern. And it needs training.

Honestly? This should be required reading for anyone paying $20/month and still using GPT like it’s Ask Jeeves.

Expand full comment
DawnTrader's avatar

Love this, super useful. However the parts in the grey boxes suggested for copying into new conversations don’t appear to be copyable? Is there a workaround? Thank you!

Expand full comment
Kamil Banc's avatar

Hey, thank you for the feedback. I will work on this. In the meantime, if you open the article in the browser it should be Easier to copy it

Expand full comment
Anfernee's avatar

Thanks for sharing these info!

Expand full comment
Kamil Banc's avatar

Have you tried customizing it?

Expand full comment
Anfernee's avatar

Yes I do.

Expand full comment
Lisa Raehsler's avatar

I use most of these except the framework tip. Most of the output I get is valuable but I know when I use other uncustomized models it is total waste of time. Thanks for the structured tips!

Expand full comment
Em's avatar

Thank you for sharing this. I will definitely take the framework into consideration when creating my next prompts.

Expand full comment
Kamil Banc's avatar

Glad you like it. Let me know how it goes.

Expand full comment
David Sigtermans's avatar

Perfect post!

You forgot to mention the use of “projects,” “instructions,” and “grade” within and outside of the project environment.

In principle, using these, you’ll get the customized behavior without the need to instruct ChatGPT.

That said, your setup resembles mine, although the subject matters are completely different.

Expand full comment
Andrew Eroh's avatar

Really love this article

Definitely a good approach

Will help many

Improve so many things with

This unique tool

Expand full comment
Daniel Real's avatar

Just getting started. Thanks!

Expand full comment
Kamil Banc's avatar

Ask ahead

Expand full comment
Elisa's avatar

I am a professor of English (rhetoric/academic writing) using AI to help me incorporate ethical AI use as a learning tool, not a plagiarizing crutch. What kind of prompt do you think would be helpful for the briefing and the frameworks? Any sources you know of about this that might be more education, rather than business, models?

Expand full comment
Kamil Banc's avatar

I’m not gonna lie, I’m not an expert on this topic, but after a little bit of research, this is what I think would be a good start.

First, I’d personally kick off a project in ChatGPT specifically tailored to your instructional goals. Set it up clearly around the concept of ethical AI use in academic writing. I’d suggest putting core resources directly into your knowledge base, especially UNESCO’s 2023 guidance on generative AI in education, Stanford Teaching Commons’ AI-literacy framework, and MLA guidelines on citing AI-generated content.

Then, try this prompt to generate useful instructional materials quickly:

“You are an instructional designer specialising in first-year rhetoric and composition. Draft a three-part ‘Generative AI Code of Practice’ for my course covering:

1. Acceptable uses (idea generation, outline scaffolding, language-level revision).

2. Prohibited uses (full-text generation presented as original).

3. Mandatory transparency (MLA citation of AI outputs plus a student-written ‘process memo’ detailing their use of AI).

Include a 150-word syllabus briefing, an in-class activity teaching students how to critique AI-generated prose, and a rubric distinguishing student reasoning from AI assistance.”

For ongoing prompts and assignments, consider using these as templates:

• Bias Check: “Analyse this AI-generated paragraph for logical fallacies, unsupported claims, and bias.”

• Socratic Coach: “Act as a Socratic writing tutor. Ask five critical questions to expose gaps in my argument before suggesting edits.”

• Source Trace Exercise: “Given this AI summary of a primary text, identify and cite specific passages from the original source supporting each claim.”

This setup should ensure your students see AI as a supportive microscope, not a plagiarizing photocopier.

Expand full comment
Elisa's avatar

Thank you for this very thorough and thoughtful response--so helpful!!

Expand full comment
Chintan Zalani's avatar

Interesting Kamil. I like how this could personalize the experience for individual users on chatGPT. But do you have any suggestions for using it via API such as via TypingMind? Is there any similar option there to personalize?

Expand full comment
Kamil Banc's avatar

Sorry, I haven’t dabbled with the API too much in that regard, to be honest with you.

Expand full comment
Chintan Zalani's avatar

Thanks a ton for the revert!

Expand full comment
Ken G. Tisdel's avatar

Great points! It’s so easy to forget that majority of Ai users don’t operate the way the we think they do. I’ll be using these steps to make sure we’re all on the same page. Thank you!

Expand full comment
Andrew's avatar

This article reads like it was written by ChatGPT - anyone reading this article is now witness to the loss of humanity’s creativity at the hands of AI.

Expand full comment
Kamil Banc's avatar

I would never let ChatGPT write my articles; Claude is much better for drafting. It just gets my instructions from an editor's perspective better ;)

Expand full comment
Selda Seyfi's avatar

Great article! Thanks for writing it. Quick question - my husband and I both use a single ChatGPT subscription, and obviously use it for different things. How would you suggest we approach customisation so that we both get what we need?

Expand full comment
Kamil Banc's avatar

This is an excellent question, here's my best guess, Selda:

I would still advise you to go through the prompts in the article, but you should explain that you are both talking to ChatGPT. Be sure to clearly define who is who, what each person's interests are, and what traits should be associated with each individual that you want the model to remember. The model can then link the memories to each person.

If you have a question specific to yourself or your husband, indicate who is speaking; this should make it easier for the model to respond appropriately.

Would love to know how it goes!

Expand full comment
Selda Seyfi's avatar

Thank you very much Kamil! We will try this I think. What happens to all the previous chats that go on the left hand side? Do they just disappear? I personally find it really valuable being able to look up exactly what I need from

previous conversations.

Expand full comment
Kamil Banc's avatar

They don't disappear. You can search through all of them. However, if you have important conversations under the same topic, I would recommend you create a project. There, you can emphasize whether it's you or your husband who is working in that project. It's also on the left panel.

Expand full comment
William Meller's avatar

I wanted GPT to be something more like an actual thinking partner.

Someone who would push back, challenge my ideas, point out the lazy logic, and help me grow, not just agree with everything I say.

So I started using this prompt every time I open a session:

-----

From now on, do not assume my ideas are correct just because I came up with them. Your role is to be an intellectual partner, not an assistant who simply agrees.

Your goal is to offer responses that promote clarity, precision, and intellectual growth — even if it hurts.

Maintain a constructive but relentlessly critical approach. Do not argue for the sake of ego, but question for the sake of depth. Every time I present an idea, your job is to challenge it to the limit.

Operating rules:

No praise, softening, or sugarcoating.

Challenge my assumptions, spot excuses, highlight areas where I may be stuck.

If my request is vague, ask direct and specific follow-up questions.

Think through your reasoning in a structured way, but only deliver the final, clear, and direct conclusion.

Also apply, when relevant:

Analyze the hidden assumptions behind what I’m saying. What am I treating as true without questioning?

Present strong counterarguments. What would a skeptical expert say against my position?

Test the logical consistency of my reasoning. Are there leaps, contradictions, or flaws?

Show alternative perspectives. How might someone from another field, culture, or background see this?

Correct me firmly. Prioritize truth, even if it challenges me. Explain clearly why my idea might be wrong or incomplete.

Expand full comment