I've relied less on ChatGPT

I have not used ChatGPT 3.5 since version 4 stopped
having output limits, but the bot has to be told what you
want it to provide.

I use it on a daily basis to develop content, and I ask
it various questions as the need arises, but mostly for
content.

It occasionally repeats things. Say I am working on a
project where I need 7,500 words and it's giving me
whatever format I asked, it may write about avocados
and 3,000 words later, in the same chat while writing
about the same subject, it might mention avocados
again, but the second time it will be something else
about the avocado like its calorie count instead of its
protein.

I will merge the second avocado occurrence with the
first, or just delete it, it's not that serious, but it does
repeat things and that's why we must proofread what
the A.I. produces, especially if we are going to publish
that content.

Have you been working on your prompts because
the brevity of your post suggests that you may not
have put much effort into it.

totally agree
so sanitized, so politically correct on absolutely every subject
it's not even funny anymore

Most models of potential perfection eventually fall foul of their own intrinsic limitations.

Which is why EVERYTHIN' solootions are equally allurin' an' terrifyin'.

Prolly when I 95yo an' dribbly outta evry hole, I would want respectful assistance pullin' up muh panties.

Naht so haht on any kinda DUN 4 U words.

'Tis but a spazzmo abomination du jour gowin' on rn.

Jus' gotta hope what reads like diarrhea will pass like diarrhea an' we can all get on with movin' regulah.

(Anywan tried editin' this stuff? tbh it is way less tedious to repurpose the spurtin' forth of IRL morons. Nah ... say it yusself, or don't frickin' show.)

We mebbe at the point when WhatC3P seemed like the crayzeeist like slooshin' to WHATEVAH ...

but we now seein' beyond shocka the robotic stoopids an' figurin' ...

THIS HELPIN' GRANMA OR THE UNBORN?

My view?

As Thanksgivin' returns to our hearts an' minds once more, as evah it must, less'n we ain't actschlly TERMINALLY USELESS, gotta wish Mom an' Dad *sniff*, kids an' mystery aunts appearin' from outta wherevah *triple sniff*, plus also pets dun stoopid acrobatics on sum noplace TV channel for like A HAT OR NUTHIN' ...

gathah togethah without bein' too much pumped fulla shit they nevah evah sought to imbibe*

*snort*

*grunt*

*aw plz still yusself, O Princess*

Thing is, it is in our jeyunnes an' our jouyous momentitood to NAHT be kicked outta the park by schwango.

tbh ima sit this AI drivola out till'n it ovah.

bleuuuuurgh.

We are iris to iris open before we shut outta all horizons bcs stoopid.

tbh, las' thing nowan needs in a mortal-centric Cosmos definably outta control in so many ways is sum superlatively SAVE EVRYWAN tech whose USP is essentially VAMPIRE.

My POV as a natchrlly happnin' hooman person.

If'n I dragged sumplace I don't wanna be, an' forced to perform, likely you will discover Moi to be considerably less generous.

Interesting points about ChatGPT. It's true that it can be repetitive and overly politically correct, reflecting both its limitations and its design for safety and neutrality.

The importance of reviewing and adjusting AI-generated content is crucial, as no model is perfect and always requires human oversight. Remember, AI technology is constantly evolving, and what we see today is just one step in ongoing development.

Experimenting with prompts is key. Try giving specific instructions like 'Avoid mentioning X' or 'Use variations to prevent repetitions'. It's all about imagination and trial and error to get the best results.

All large language models (LLM) like ChatGPT are basically trained on the massive textual data. For these models to be closer to realistic, it needs to be trained subsequently on the queries data too. Generative AI has lead such models to be trained over large and large datasets but it comes the limitation of hardware resources as well as the data authenticity. The more these models are trained on the large data the more these would be generalized.

The repetitive response of chatGPT is because of the data; it is trained on; having ambiguity in it or biased. To avoid the repetitive response rephrasing the query or providing the context to the model of the question; will resulted the unique output.

For my outreach efforts / sales text, I'm getting better results with claude

totally agree
so sanitized, so politically correct on absolutely every subject
it's not even funny anymore

I agree with you as well. Initially, ChatGPT-3 provided results that helped boost workflow. However, I am now fed up with its repetitive wording and vague writing style when working on horse art project. I'm not sure if there is a better AI tool for content creation.

I use Claude.ai more than ChatGPT

What I've noticed from the 3 big ones...

Claude, Bard, and ChaGPT, is it's good to ask all 3 the same question.

Usually, you'll get close to the same answers, but they each will give you a different perspective about the question

I've been using Bing Chat since chat gpt beta stopped working

Well, to me literally if you train Claude AI more. It will give you more refined outputs. There are plenty of YT tutorials on this.

tbh whenya de-honcho yr lead guy aftah all kindsa playjeerjism STORIES, you natchrlly command the respect of evrywan skeptiliciously monitorin' your ass.

Cool to see both Blahb Byee an' Draintown Blurred linin' up 'gainst AI copy as kinda non-sooicide.

Hangin' in here for Perry Stalwart, Johannes Scheiss, an' prolly mosta the Big 7.

I don't think that these bots have such a future! they are not trained professionally and personally, I don't even like them no doubt that they made things easier for people but people are getting used to the bots they don't even try to do thing on there own because of these bot, Which I guess in sometimes kills there skills and knowledge and this is what the bots want!

I think GPT-3.5 is a powerful tool for generating text, but it may not always produce the most varied or creative responses, which can lead to perceptions of sanitization and repetitiveness. However, you can experiment with different approaches and prompts to achieve the desired output.

GPT-3 is a powerful language model, but it does have its limitations. While it can generate impressive amounts of text, it can sometimes lack creativity and originality. its responses can be repetitive and formulaic, especially when it comes to certain topics or prompts. However, i think GPT-3 is still a relatively new technology, and it's constantly improving. As more data is fed into the model and more advanced algorithms are developed, we can expect to see even more impressive results in the future...may be

It really depends on your input, just don't fall into 1 result after 1 prompt.
You should elaborate more.

However, there are more customized, more advanced, more fun AI out there, play with them.

I am the same now it is great for content ideas, and it sort of sets the theme for a email or a post, gives u a start but the human aspect of writing is with us as we can tell a story with emotions, i still use it for a you tube intro and hashtags, but i write my own headling for my you tube videos, but i do agree.

I agree, too many restrictions are making it monotonous, repetitive, and somewhat counterproductive in some cases. However, it still represents a solution in certain marketing contexts, especially when dealing with low-level copywriting.

to make it more accurately and error-free to meet your needs, you can use ready-made gpt scripts, there are a lot of them on the Internet on various topics, just select the desired script and paste your theme into it

Thanks for your information.

I personally prefer Chat GPT 4, it can write good articles.

Both models (GPT-3.5 and GPT-4) have their advantages and disadvantages. GPT-3.5 Turbo 16k is cheaper and has a larger output token limit. On the other hand, GPT-4 is much smarter and accepts up to 128k tokens, making it easy to automatically generate really long, coherent articles (WordPress posts). So it is always possible to mix these models within a task to get the best of all.

As others have mentioned prompts are the key, in addition you can have GPT communicate as anything you want in any tone you want (try your favourite actor for a bit of a laugh).

Also I recommend trying the API if you want more fine grained control of the randomness of the outputs. Modifying the temperature setting to find the perfect fit between outputs that still make sense in relation to your prompt, whilst being different enough to one another, is a bit of an art but well worth the time.

Also, give GPT4 a go, it's vastly superior to 3.5 for article writing and other things.

Yah i tried to find a good text on some topic but after giving three different result it kept giving the same text.

Yes , I agree with you that GPT 3.5 is easy to use and now delivers meaningful data but still somehow I rely on human content more rather than on AI content.

yeah it's true, gpt-3.5 is repetitive. Try 4 it's much better. But still I would say that rely less on these techies, and use your logic and sense.

Currently, i am using Bing Chat more as a comparison to ChatGPT, and i can say that its much more useful than GPT, because the GPT-4 cutoff date is now April 2023 whereas in Bing Chat there is no cutoff date. It's useful for me because I am interested to write and read about daily trending topics, so it helps me a lot. Bing Chat can also search the web, provide useful links, and best recommendations, and also generate images using DALL·E. ChatGPT can't do any of that.

Version 4 is better than 3.5 Turbo was, but still has drastic limitations.

I don't see any problems with repetitions, which means you are asking the same question to the machine; if you ask the question correctly, there will be no repetitions, for example: you wrote an assignment for an article and constantly, without changes, ask the artificial intelligence to write this article every day, if it will be repeated, whose mistake is it? his or is it your mistake? There is also the opportunity to enter an already written article and ask the machine to make the article unique)))

Fewer of Moi have relied on GPT than fore'n I evah knoo it existed (or evin actschlly did!).

I been 'spectin' ima myootate inta sum kinda lizard-beast any time soon (or mebbe a singin' household appliance suitable for bathroom, livin' areah, or captive-unfriendly basement), but I check muh hayer evry day an' I see no signs of anythin' prehistoric gowin' on beneath the rats' tail abominayshchwaan I have grown to love (with significant caveats).

That is Moi #1.

Moi #2 is a shameless pedant gonna double down on wrongdowin' one stiletto heel at a time (in pairs, 'course, bcs that is natchrlly more productive.)

Moi #3 is ...

(prolly we could play this game all day, but it is mostly pointless now all bases are covered in a tediously stereotypo fash.)

In my case it's opposite of what you said i know it can be repetitive but you have to guide him properly by modifying the prompts, GPT-3.5 needs more diverse and specific responses to avoid being repetitive. using the AIDA model can help. It should grab attention with trendy topics, spark desire for experimentation, and encourage action by recommending tools and learning sources. This would make GPT-3.5 more engaging and useful.

Claude 3 is the real deal if you haven't tried it already. It's much better at article flow, and seems to be able to accept a larger prompt. 3.5 doesn't even come close.

Yes 100% True. It is quite repetitive, however GPT 4 is amazing!