A good test run is to list the ingredients you have on hand and ask for a simple recipe, or a meal plan for the week. We’ve also used it to assist with travel planning. It’s so helpful.
2 Likes
I’ve used it to create form letters…BUT I have always proofread and edited what was generated.
The use of ChatGPT should not take away the need for thinking and making sure what it generates is what you really need and want.
1 Like
Keep in mind that with the free version of ChatGPT, there is a limited number of questions you can ask per day. And typically, one discussion can involve quite a few questions as you refine what you need. It’s frustrating to get part of the way there and then get the message that you’ve hit your limit. I bit the bullet and paid for a subscription.
Yes. We’re finishing out the back half of our lower level to add another bedroom with built-in bunks (you know, just in case grandkids ever show up). I fed Chat a pic of the main bedroom area for decor continuity along with the dimensions of the wall we want the beds built into and included door dimensions as well. After a few iterations, I got something I liked well enough for DH to model with his CAD system.
Here’s the pic of the existing area I gave it:
Here’s what I eventually ended up with after refining my request to add lighting, drawers, ladder, etc.
It beats looking through hundreds of online photos to come up with something close but that I’d still be modifying. This image is exactly what I want and exactly what I’ll get.
7 Likes
This is very cool. I appreciate you sharing this example!
And you did this with just the ChatGPT website or app?
The app.
1 Like
This can be done on the website too, but you need to be logged in.
Great example @ChoatieMom!
I did something similar for our kitchen remodel.
1 Like
How much is a subscription? I’m finding I’m using it more and more.
$20/month. I can deduct it as a business expense, or I probably wouldn’t pay for it.
Thanks. I don’t use it enough to pay. But it’s good to know there is a daily limit.
Edit: I tried to do this in chat, but I couldn’t get it to work. But perplexity did. I have photos of clothes without stock models for Poshmark. I asked it to put it on a model. This is what I got. I get 3 free a day.
3 Likes
Interesting use, @conmama.
@conmama did you specify a size for the model? I would do that because often the things I want to buy are on a size 2 model who is 5’ 10” tall…and that’s not me!
1 Like
I didn’t, but that’s a great idea! This was just my first attempt. I have lots of plus sizes so that’s a great idea! I might go ahead and pay for one month to see if it helps sales.
2 Likes
For professors: Can ChatGPT analyze if it were substantially used in a submitted report? If it keeps a copy in the cloud of all “correspondence”, it might be able to cross-check how much of a report was generated by itself (even without any personal information).
1 Like
I asked, “If I give you a report, can you tell if it was generated by ChatGPT?”
" The short answer is: no tool (including me) can reliably tell you whether a report was generated by ChatGPT or another AI. Here’s why:
Why Detection Isn’t Reliable
- No hidden watermarking: Current ChatGPT outputs don’t include invisible tags or signatures.
- Overlap with human writing: AI-generated text often looks very similar to well-edited human writing. Style alone isn’t proof.
- Detection tools are weak: Existing “AI detectors” (like GPTZero, Turnitin, etc.) have high error rates. They sometimes flag human work as AI (false positives) or miss AI work (false negatives).
- Edits blur the line: If someone lightly edits or rewrites an AI draft, it becomes even harder to distinguish from human work.
What Can Be Done
Instead of certainty, you can only get probabilities:
- Stylometric analysis can look for patterns (predictability, repetition, phrasing).
- Comparison with a person’s past writing can sometimes highlight inconsistencies.
- Specialized software can give a “likelihood score,” but it’s never 100%.
Bottom Line
If you send me a report, I can analyze the writing style and give you an opinion on whether it resembles ChatGPT-style output. But I cannot say for sure—it would only ever be an educated guess, not proof."
3 Likes
"Here are the main features I’d look for when evaluating whether a report might have been generated by ChatGPT (or another AI). You can use this as a checklist when you look over text:
1. Structure and Formatting
- Well-organized sections: AI tends to produce very neat, outline-like structures (headings, subheadings, bullet lists).
- Balanced paragraphs: Often similar in length, typically 3–5 sentences each.
- Predictable flow: Introduction → body → conclusion, without abrupt topic shifts.
2. Language and Style
- Polished but generic tone: Neutral, professional, and rarely showing strong personal voice or quirks.
- Overly clear transitions: Frequent phrases like “In conclusion,” “It is important to note,” “Overall,” “This highlights”.
- Limited idioms or unique phrasing: Human writers often include slang, humor, or personal phrasing — AI usually doesn’t.
3. Content Patterns
- Balanced coverage: AI often tries to cover all sides, even when the task doesn’t require it.
- Repetition of ideas: Concepts may be rephrased multiple times instead of expanded with depth.
- Shallow analysis: Lots of surface-level explanation, fewer surprising insights or personal anecdotes.
4. Consistency and Voice
- Uniform style: No noticeable shifts in vocabulary or tone (humans often vary naturally).
- Overuse of certain words: e.g., significant, crucial, key, important, overall.
- Limited errors: Grammar and spelling are almost always correct — but sometimes a little too perfect.
5. What You Won’t Find
- Hidden metadata or fingerprints — nothing in the file/text itself reveals it was AI.
- Absolute certainty — even with all the above signs, it’s still only probability.
Practical Approach: If you want me to check a report, I can:
- Look for these markers,
- Compare it to known human writing (if you provide a sample),
- Give you a “likely AI / likely human / uncertain” assessment.
Would you like me to walk you through a side-by-side comparison (AI-like vs human-like excerpts) so you can see the difference in action?"
Every time I ask it a question, it ends the answer asking if I would like it to do more. That’s why it’s easy to hit the daily limit of questions! Each response by me counts as one.
1 Like
I think ChatGPT can be a great tool for use in everyday life and when appropriate, work life. But I am never going to take the info it spits out as 100% accurate. Always use some of your own brain too!
1 Like
No one should. GenAI is known to occasionally produce incorrect (sometimes completely made up) information. These are called “hallucinations”, and AI researchers are working on ways to reduce, if not eliminate, this phenomenon.
3 Likes