People were stupefied by the appearance of ChatGPT on the AI scene. Powered by the GPT-3.5 version, this AI text generator answered any questions, searched for any piece of information, and created good texts using AI essay writing capabilities.
✅ AI Essay Writer ✅ AI Detector ✅ Plagchecker ✅ Paraphraser
✅ Summarizer ✅ Citation Generator
However, the developers of the AI tool didn’t stop there. Recently, the launch of the new system upgrade was announced. The GPT-4 model first became available to the ChatGPT Plus subscribers and now it opened its access through API to outside developers. The update promised to enhance its capabilities in reasoning, following complex instructions, and improving creativity in AI essay and blog writing. But most importantly, It offers limitless possibilities and can be a rather useful essay help service.
As we already tried working with the previous model, we’ve decided to check out, whether the new one would be better at generating essays or whether it was just a matter of hype.
GPT 3.5 vs GPT 4: First View
| GPT-3.5 | GPT-4 |
| Free plan | $20/month |
Both models have the same interface as well as the general way of usage. The main difference, however, between the two best ai content generator tool systems is that the 3.5 version is free and the 4th one is part of a $20 monthly Plus plan.
The older one has its limitations, as it can only work when the demand is low. With the subscription, on the other hand, you don’t only get access to the new updates, but you also get to use the platform when there’s an influx of requests.
Testing
To put the two models in a fair game, we asked the AI essay writing generator to complete the same single set of instructions. The prompt looked like this:
“Write a 550-word personal narrative essay about a story exemplifying a feeling of gratitude”
We didn’t adjust or specify our request with any of the models. The texts were taken as they were, with no changes made.
The generated results were then organized into MLA-formated documents. We then sent them out to our partner US college professor, Sybil Low, to be checked according to all the accepted standards. The set grading minimum was 60/100 pts.
Results
Generation time
| GPT-3.5 | GPT-4 |
| 2 min | 5 min |
The first thing we noticed was the waiting time: with GPT-3.5 it took up to 2 minutes max before we got a full text. GPT-4 though took up to 5 minutes to come up with the result. Not such a drastic difference but we still worked with an updated version for a bit longer.
Generation flow
| GPT-3.5 | GPT-4 |
| Can be only accessed when demand is low | 25 requests for every 3 hours |
What’s important to note is that GPT-4 as a recently-launched AI text generator model has its restrictions. It can only process 25 messages every 3 hours. If there’s more, it can sometimes experience small bugs or failures during the generation of the full texts. GPT-3.5 also has its limitations, as it can only be used when the demand on the website is low.
Wordcount
| GPT-3.5 | GPT-4 |
| 552 | 636 |
Both GPT-3.5 and GPT-4 fulfilled the word count requirement. The first one generated 552 words, and the second one even exceeded the needed amount creating a 636-word long text. This is a good sign because it means that you won’t have to create the needed paper part by part but can rather receive a whole piece at once using AI essay writing.
Professor’s Evaluation: Essay Analysis
| GPT-3.5 | GPT-4 |
| 83.9/100 | 88.9/100 |
The grading aspect is the most important as it reflects the quality of the completed work. We wanted to evaluate both works to compare them side-by-side and find whether there is an actual difference in quality. Our main question was: is GPT-4 really better and worth paying for?
As the two papers came back after the assessment, the 3.5th one received 83.9/100 pts and the 4th version got 88.9/100 pts. This is not such a big difference, still, let’s dive a bit deeper into the results.
Both papers received 100% for spelling. The two narratives were also marked identically in terms of efficiency, gaining only 51%. Now, there were a lot of differences between the results. All of the other indexes were higher in the GPT-4th essay than in the 3.5th one. Grammar raised from 59% to 86%, and punctuation from 89% to 96%.
There were also improvements in the reasoning and logic section: acuity went up from 63% to 82% and objectivity from 74% to 94%. Some of the aspects, though, faced a bit of a dropdown of points. The word choice, for example, in the GPT-3.5 text was marked with 96%, while the last model’s paper gained 82% here.
Our partner professor Dr. Sybil shared some notes regarding the papers:
“I’ve noticed some similarities between the two works, however, essay 4.0 was a bit better. I liked the structure and the different parts of the text seemed more connected. I would say that the second work was more thought-through and therefore organized. And it was for sure more pleasant to read.”
Here’s a more detailed breakdown of both grades:
| Essay by GPT-3.5 | Essay by GPT-4 |
| “SCORE 83.9 out of 100 pts. Document formatting is very good, but there is room for improvement. Document formatting issues: vertical spacing; page header. {Integrity}: 100%. {Length}: 100%. {Mechanics:} 86% (spelling 100%, grammar 59%, punctuation 89%, word choice 96%). {Citation formatting:} ungraded. {Reasoning, logic:} 72% (efficiency 51%, acuity 63%, clarity 100%, objectivity 74%).” | “SCORE 88.9 out of 100 pts. Document formatting is very good, but there is room for improvement. Document formatting issues: page header. {Integrity}: 100%. {Length}: 100%. {Mechanics:} 91% (spelling 100%, grammar 86%, punctuation 96%, word choice 82%). {Citation formatting:} ungraded. {Reasoning, logic:} 79% (efficiency 51%, acuity 82%, clarity 89%, objectivity 94%)” |
It seems that GPT-4 mostly showed an improvement in scores and quality of the AI essay writing content compared to the work presented by its previous version.
Essay by GPT-3.5
Essay by GPT-4
Plagcheck of the essay by GPT-3.5
Plagcheck of the essay by GPT-4
The Verdict
We’ve created two essays: one using the GPT-3.5 free version, and another with the help of a recently updated GPT-4 AI text generator. We put the works back-to-back to find out whether this new model is better than its free variant in AI essay writing. And, as it seems from our experience, it does beat GPT-3.5 by a few scores. As was promised, the new model improved the reasoning of the generated writing. From our judgment, it also made a more interesting and well-written narrative than before. Hence, we would say that GPT-4 does provide improved results in terms of text generation. The quality gap is not big, but it is there.
FAQ
Can I use GPT-3.5 for free?
Yes, GPT-3.5 is a free version of the AI text generator. However, it has its limitations and can only work when the demand is low. If you subscribe to the Plus plan for $20 per month, you not only get access to the new updates but also get to use the platform even during high-demand periods.
Are there any limitations on using GPT-4?
Yes, GPT-4 has certain restrictions. It can only process 25 messages every 3 hours. If there is a higher volume of requests, it may sometimes experience small bugs or failures during the generation of full texts.
Did GPT-4 show an improvement in quality compared to GPT-3.5?
Yes, according to the evaluation by a college professor, GPT-4 showed a slight improvement in quality compared to GPT-3.5. The professor noted that the structure and connectivity of the text in GPT-4 were better, making it more thought-through and pleasant to read. However, the quality difference between the two versions is not significant.
Is GPT-4 better at maintaining word count requirements?
Both GPT-3.5 and GPT-4 were able to meet and even exceed the word count requirement in our test. However, GPT-4 generated a longer text overall.
Is GPT-4 worth the extra cost?
While the quality difference between the two versions isn’t drastic, GPT-4 did receive a higher score in our evaluation and was found to produce more coherent and well-structured essays. Therefore, for those looking for better-quality AI-generated essays, the upgrade may well be worth it.
Follow us on Reddit for more insights and updates.
Comments (4)
Welcome to A*Help comments!
We’re all about debate and discussion at A*Help.
We value the diverse opinions of users, so you may find points of view that you don’t agree with. And that’s cool. However, there are certain things we’re not OK with: attempts to manipulate our data in any way, for example, or the posting of discriminative, offensive, hateful, or disparaging material.
More from Academic Writing Tips
This is crazy! Both AI models seem impressive, but I can’t help but wonder if we’re getting lazier or smarter with these techs.
Wait, so GPT-4 is better at reasoning but worse at word choice? Hmm, now that’s a conundrum… 🧐
Good catch! Maybe it’s more sophisticated but less creative? Can an AI have a “style”?
I have been using GPT-3.5 for my essay drafts, and it’s been pretty helpful! Might have to consider this new GPT-4.