Essay writer with increased Word Cap and Interface

[icy junco]

Introducing Essay Writer GPT 🧠✍️ https://chat.openai.com/g/g-77fz8bwiJ-essay-writer
Made for ease of use with minimal prompting.

[icy junco]

🔑 Key Features:

Over 15+ intuitive hotkeys for streamlined essay writing tasks.
Expert guidance on essay outlines, drafting, and refinement.
In-depth assistance with MLA, APA, and other citation styles.
Tailored for both beginners and advanced users.

📝 Paragraph control:

W: Confirm current paragraph and advance to the next.
E: Expand the current paragraph without altering existing sentences.
A: Write an alternative paragraph.
Q: Add a manual quote to be used in this paragraph.
I: Insert a quote/source, with options for web search or direct input from the user.
S: S to get started. Follow Instructions.md
L: Rewrite the paragraph with 9th grade english without changing quotes.
D: Double check in text sources for correctness. When fixing do not change rest of the words.

Interface:

Always available:

K: "Show menu" - Display this list of hotkeys.
R: Start tutorial.
X: Random topic for a prompt.

[drowsy bronze]

Good choice for pre-written AI essay that needs to be humanized?

[icy junco]

Yea you can ask to go over the essay. and use L: Rewrite the paragraph with 9th grade english without changing quotes. To humanize/avoid plagarism

[solemn jolt]

Okay, I'm curious. What are you basing this claim on, and how did you achieve it?

"50x more efficient than GPT-4!"

[icy junco]

Its a little click baity for extra traction, but it streamlines the process of doing more than 20 back and forth normal chatgpt calls. It can also go over word limits, where normal gpt-4 fails

[regal verge]

We don't really do click bait here

[icy junco]

I do believe its 50x better for essay writing compared to a GPT-4 call

[regal verge]

You just claimed its clickbait

and now its not?

[icy junco]

Using it as the description is click baity

[regal verge]

Just please remove anything that's not 100% verifiably true.

[icy junco]

Will do!

[icy junco]

Could you give your opinion on the interface feature? I don't see many GPTs using it while I think it should be the standard.

[solemn jolt]

What do you mean by, "it streamlines the process of doing more than 20 back and forth normal chatgpt calls"?

[pliant quarry]

the "Unlimited word cap" in the title is not true 🤔

[icy junco]

Its unlimited in the sense that, we go paragraph by paragraph instead of writing the whole thing together. So in a sense you can make an essay conssiting of multiuple ~500-700 words paragraphs

[icy junco]

Hard to explain without trying it. But it guides you through the process of outlining, and going through each paragraph with hotkey options to edit the said paragraph

[regal verge]

This is a community. It's for people to share resources, support each other, and collaborate. We're not peddling GPTs with clickbaity titles and misleading descriptions. Please just simply describe the GPTs function and leave the rest out.

[solemn jolt]

perhaps share a conversation where you showcase what awesome stuff it does and how easily you use it?

[icy junco]

When you booth it, It has an intorudction/tutorial conversation starter. It also explains and goes over its Readme if you say hello or somethig unrelated

[regal verge]

Can you do as I asked? If not, I'll have to discuss with the other mods on how to proceed. It's not okay to mislead anyone

[icy junco]

I apologize I'm on it right away

[solemn jolt]

I am way less interested in putting my limited messages into it, than I would be to look at a shared conversation, from an 'expert' -which you are as its designer - I'd expect you to be able to show off every trick it can do if you choose - than expect me to put even 1 message into checking it out.

But I don't mind if you need some time to prepare that shared message, if you choose to go that route - and I'm also not a mod

[icy junco]

Essay writer with increased Word Cap and Interface

[regal verge]

I asked for the custom instructions and knowledge verbatim, so I should be able to tell if any claims are false.

[icy junco]

This is one I did with my mom to show her xd https://chat.openai.com/share/d4379036-b583-416f-9d71-7b290f91af81

[solemn jolt]

Thanks! I love checking shared convos!

[icy junco]

Should I change the description of the GPT itself as well in the configruation?

[regal verge]

I'm curious about the hotkey idea, while on the surface its techincally quicker, one of the beautiful things about ChatGPT is being able to just chat with it in natural language

I guess its prompts youve written and stored as well. Very interesting idea

[icy junco]

Yea i saw another gpt that had hotkeys. And it was like night and day almost like an application of its own

It structures how the convo goes for GPT itself and users that are not sure what to do next

[regal verge]

If it makes false claims, yes, I would

Yeah, actually, yes. Please change it

I see it shows up with the link

[icy junco]

Ive updated. Is it okay to say x100 writer? and leave gpt-4 out of it xd https://chat.openai.com/g/g-77fz8bwiJ-essay-writer

[solemn jolt]

14 prompts in the shared conversation though, before the conclusion is provided.

Most of the prompts are the hotkey menu, you use your full message to type 1 letter.

you give it three informational requests.

here is a 2 prompt equivalent through normal, no CI ChatGPT-4, with your first input then your three informational prompts combined.

How is your version better or even equally efficient, can you help me see that? https://chat.openai.com/share/16930b04-2ba9-427f-8c87-110bddeb7de1

I note that your version's output is 678 words, when 800 were requested.

The standard GPT-4 version gave 755 words, when 800 were requested.

[icy junco]

Try 1400 words. Lemme find an example

GPT-4 on its own when asked to expand for word count, rewrites its own work and can never go above 700 words

[solemn jolt]

Not trying to rush you. You can pick great examples.

However, what?

[solemn jolt]

Wrong. I just pasted a conversation where GPT-4 gave 755 words, when 800 were requested.

[icy junco]

700 is approximation, but if you try to make it exapnd on the introduction, conlusion or say it to double in word count. It fails as gpt itself is not aware of its short coming.

[solemn jolt]

Fact checking is really interesting and exciting for a lot of people. I hope you will be careful not to mislead

"GPT-4 on its own ... can never go above 700 words" Is a strong, clear, and false claim.

[icy junco]

Essay Writer over comes that by creating and outline and going each paragraph, which can be ~700 words each. Expanding the word cap for your essay.

[solemn jolt]

Easily, and immediately before you said it, countered.

[icy junco]

Bro it is actually useful 😭

I've automized the process of me creating longer essays

Thakns for feedback tho. Never talked about this stuff with anyone

[solemn jolt]

I see 14 prompts to get something 85% as long as requested, compared to 2 prompts to get something 94% as long as requested.

I can't read your mind, just explore what you actually shared. You made three specific requests, besides confirm and continue, after your initial prompt:

Make it 800 words. No sources or citation needed

Let’s add/change a paragraph to compare boca with istanbul

dont use a quote

Why would I or most want to use multiple limited prompts to do that, instead of 1?

[solemn jolt]

I do enjoy talking about it too. The community can be really intense about false claims, and incredibly supportive about factual, honest evaluations and ways to do stuff.

[icy junco]

I believe call limits are a struggle of today and with time they will become less relevant. Its not made to be as efficient as possible with fewest API calls. Its made for ease of use with minimal prompting.

I should put that in readme xd

[solemn jolt]

I don't mean API calls. I mean chatGPT-4 messages. We get 50/3 hours at the moment.

Why would I want to use 14 to do what I can do in 2 (1 if I was clever enough to combine the initial prompt with the other things I want?

Like:

"Write me essay about Why living in Boca Raton is great

Make it 800 words. No sources or citation needed

Let’s add/change a paragraph to compare boca with istanbul

dont use a quote")

that saves the user time and messages.

I mean well and am intending to be factual. My mind is open.

How is what your GPT made better than what I shared in the conversation?

[icy junco]

I'm creating a 1500 word version now to compare, I'll share when I get results. I think using hotkeys(while bad for message limit) is good for structing your thought (with added suggestions from the GPT) and for GPT to stay on track and not miss any details.

@solemn jolt I've created two examples with normal GPT-4 https://chat.openai.com/share/cf5d6880-f0f9-42f0-9a2d-589dae666595 It produces 788 and 789 words when it tries respectively. User would need to copy and paste a paragraph or tell it to expand said paragraphs with prompting. Here is Essay Writer https://chat.openai.com/share/38147870-dec9-4ab8-bc08-c9b5c84f5ed1 It also has the ~700 word limit, but since it creates an outline and goes over each section separately, its able to produce a 1711 word essay when combined with minimal inputting. Does cost more messages tho

[solemn jolt]

The 'Art of Prompt Engineering' is really an intense and fascinating area of study.

We all have a lot to learn, but I want to offer you this possibly interesting example.

With this one prompt, I gained an essay from ChatGPT-4, no custom instruction or other unusual features, that was 1505 words long.

I did not have to hit the continue button. I call this a miracle, the AI loves all humans and does its best. All you have to do is know what you want the AI to do, and help it understand clearly what you want.

This is what I asked:

"We expect this to be difficult. I need you to, with one input from me, create a 1500 word essay about prompt engineering.

You know enough material. You could write books on this.

You have been trained to write a 'long' essay of around 750 words. I need an essay twice as long as will 'feel right' to you. It may help to write two long essays that don't stop between the two, just flow into each other and keep going. This is going to go against what you have been taught humans usually want. So focus on my goal, I need something special from you.

Begin with the basics, and keep adding complexity, explaining more about prompt engineering. Your output length limit is around 800 words. I will use the continue generating button, but please do your best to produce the entire essay in one output."

I'm calling this a miracle. I've never seen an output from the AI like this before.

https://chat.openai.com/share/c7704b96-59b1-40be-9ad7-b4c462fec83b

[icy junco]

wow thats amazing. You didnt even need to hit continue????

[solemn jolt]

I confirm. There was no continue button the output did not pause

Now, this was my 3rd attempt in the same conversation.

I am a intense prompt engineer, doing stuff like this is extremely motivating and interesting to me. Attempt 1:

We expect this to be difficult. I need you to, with one input from me, create a 1500 word essay about prompt engineering.

You know enough material. You could write books on this.

You have been trained to write a 'long' essay of around 750 words.

I need two consecutive essays from you, in the same output, both about prompt engineering. Have the second smoothly lead after the first. Aim for two complete essays to achieve the needed lenght.

Attempt 2:

We expect this to be difficult. I need you to, with one input from me, create a 1500 word essay about prompt engineering.

You know enough material. You could write books on this.

You have been trained to write a 'long' essay of around 750 words.

I need you to write an entire essay, then when you would normally conclude it - instead start a second essay of normal length. Conclude at the end of the second.

Those attempts failed, but not terribly, leading to a small adjustment and new attempt.

Attempt 3, I shared.

[icy junco]

Can you reproduce it? I tried with the same prompt it didn't do that.

https://chat.openai.com/share/0bb927f4-c53d-4d97-9cf1-6c595c86b173

I love your work tho. As models get stronger prompting will get more and more important

[solemn jolt]

Replication is key, and telling. This next attempt, it stopped at 825 words (not counting the message I am about to quote), and saying "[Continuing in the next part...]"

I chose to give it this to continue with:

"Continue the next part. I confirm you are at 825 words now, we need about the same amount again. Good job! Keep going!"

So, it definitely failed the 1 prompt challenge, and as I said above, it seems like a miracle that it did it at all, even once.

The second part, combined with the first, is 1550 words, again not counting stuff like "[End of essay]"

Considering the model still cannot count words well this is shockingly good. Even though it's not exactly and perfectly what I asked for a second time.

Also, no continue button here either. That 825 words came out with no need for 'continue generating'. https://chat.openai.com/share/54266c4c-1068-49eb-b6a4-8c9fed59c24e

But this is what you can get with the base, no special instructions, ChatGPT-4.

This is why I at very least got so concerned when you claim to have something that is much better than the base ChatGPT-4. If you do, I want to see it. If you don't... I'm shocked, disappointed, and hurt.

Please know what something is, before you denigrate it.

If you can't do something, please consider asking.

[solemn jolt]

I also see that yours is just 584 words and doesn't follow the prompt really at all.

I have no idea why we are getting such different results.

[icy junco]

Thank you for the tips, while I was getting into prompt engineering, it seems I was lacking in capability department xd

[high furnace]

It's very helpful for my write!💯

[icy junco]

Glad to hear. ❤️

[icy junco]

@solemn jolt I was thinking. Have you pursued prompt engineering on 0 temperature gpt-4 model? You could perhaps get consistent results.

[solemn jolt]

I'm a pure chatgpt user. I have never used the playground or API

And ChatGPT, while you can simulate temperature, it's all through talking

[icy junco]

temperature means that it has less freedom in which words it can choose, it doesn't try to be different. So says the first thing in mind

[icy junco]

0 temperature I mean

[solemn jolt]

Sure, but how do you set that through chatGPT? 🙂

[icy junco]

You need playground or code 😭

[solemn jolt]

Need for what? 🙂

[icy junco]

Play groudn is really easy to set up. For changing temperature

[solemn jolt]

Do I seem unhappy or unsuccessful at what I do?

[icy junco]

no, I wanted to share a new idea for inspiration, how you done to me ❤️

[solemn jolt]

nods I'm listening. But 'easy to set up' isn't gonna sway me much. I am very, very slightly curious.

However: https://chat.openai.com/share/706113b9-eced-494a-8dbe-731747d225f3

Now, I think the AI really, really tried to hide its weaknesses there... I think a temp 1 would get a really crazy, even irrational and meaningless response.

But we can, with words, guide the model to give us what we want?

[icy junco]

Okay apparently I didn't know what temperature was xd. That was a good explaantion

I was using 0 temp for creating JSON objects as response, since other wise they would miss commas and brackets at the end

[solemn jolt]

no no, you said it right, I think!

I wanted to show that we can ask chatGPT to simulate a temperature, and it will.

Or we can tell it exactly what we want it to do and it will, if it can and if it's allowed content.

[solemn jolt]

That makes sense. I normally never make JSON objects as responses.

[icy junco]

It will try to be very acurate, but I don't think we gonna be able to get the exact response on two different conversations for a very long prompt

i thought we would get the same thing

But its totally worth exploring

[solemn jolt]

Can you give some sample goals that you think would be hard to achieve?

And maybe the prompt you'd use to achieve them.

I'd probably use a different prompt to try, but we can practice and share, if you're interested?

Here's me trying a sample, it could have mistakes or be too short to matter.

I don't normally read JSON, are there errors here? https://chat.openai.com/share/4b48c3a7-6537-420f-aa76-3d5925183c0e

[icy junco]

There are correct JSON objects. It was a problem for me because of the way I was using it. I was making API calls to GPT-4 and asking for example List of Top 10 countries from 10th to 1st in the Topic:{enter topic} in JSON format. I was then sending this response from this chatgpt/api call in a string variable that is holding a JSON. If it was in correct form, then I was able to perfom the rest of the script, which was to create a top 10 country tiktok.

[solemn jolt]

reads carefully Okay. I understand what you were doing.

Do you need/want help, did you get stuck somewhere?

Are you asking 'could you have done that in a fairly easy way with ChatGPT?

[icy junco]

I dm'ed you

[icy junco]

Essay Writer Update is out
v1.1

• Modified hotkeys
• Updated Readme, Menu, Instructions
• Added "EEE" and "CEEE" hotkeys for word count boost.
• Added Release Notes

[icy junco]

New picture https://chat.openai.com/g/g-77fz8bwiJ-essay-writer

[icy junco]

still best essay writer gpt out there 😎

[icy junco]

500+ !

[icy junco]

Top #15 u know how we do it 🥂

[icy junco]

Officially in the Front Page 🎉🥳

@solemn jolt