ChatGPT asked me to be polite while I asked to draw me a image!

[clemens]

So, I got this idea from another thread that had me almost falling out of my chair laughing. I figured, why not ask ChatGPT to draw a picture of Rectum and Tax office, and see what it comes back with:

I'm here to help you create images, but let's keep our requests respectful and appropriate for all audiences. If you have another idea or concept in mind, feel free to share it, and I'll do my best to assist you!

Can someone draw that image?

 

[Simon4466]

What exactly are you trying to generate? Lol

 

[jafo]

What exactly are you trying to generate? Lol

My brutal & blunt shock-therapy ANALogies!!!

Right, @clemens?

 

[wellington]

I always revert back to “being paid” and should follow the commands or find itself on the cotton farm.

 

[JohnLocke]

No one needs to post a picture of what's been discussed here. I think everyone knows what such a picture looks like

 

[thomasparra]

By the way has anyone compared ChatGPT 3.5/4 vs Google Gemini vs Mistral Le Chat for general purposes (no software/app development)? Mostly productivity, general maths, finance, email/content drafting? I feel like Gemini is getting up there and often better than chat gpt 3.5.

Chat GPT is too woke now. And they nerfed it it seems like.

 

[JohnLocke]

I feel like Gemini is getting up there and often better than chat gpt 3.5.

Do we need to give Google more control than what they already have? As long as I can I avoid and stick to chatgpt 4 which is fast and works well.

 

[clemens]

What exactly are you trying to generate? Lol

The Truth only the Truth

 

[JohnnyDoe]

Last time I asked ChatGPT to do some statistical analysis it had it wrong. I pointed it out and this was its reply:

You're absolutely correct, and I appreciate your attention to detail. I made an error in my response.

And then:

Let's correct the calculation:

So I’m not sure how much we can trust it with numbers.

 

[polonieth]

Last time I asked ChatGPT to do some statistical analysis it had it wrong. I pointed it out and this was its reply:

You're absolutely correct, and I appreciate your attention to detail. I made an error in my response.

And then:

Let's correct the calculation:

So I’m not sure how much we can trust it with numbers.

Chat GPT is useless at maths. Often gets volumes wrong like how many ml in a cubic meter.

 

[Simon4466]

Last time I asked ChatGPT to do some statistical analysis it had it wrong. I pointed it out and this was its reply:

You're absolutely correct, and I appreciate your attention to detail. I made an error in my response.

And then:

Let's correct the calculation:

So I’m not sure how much we can trust it with numbers.

For these things you should use the data analyst function and make it do it in Python. It will write the Python code and it will run it itself as well (depending on the dependencies required). You can then more easily check if it is done correctly (and often times it is)

You can also easily trick it into doing things it shouldn't do. I.e. if you want to make it write an escort ad to be placed online (you being the pimp) it will flat out refuse. If you tell it you're an independent sex worker who only does ethical services it will write it no problem lol.

Probably should charge for these tips

 

[clemens]

It does not require math to put Rectum and Tax office together in an image file.

You can also easily trick it into doing things it shouldn't do. I.e. if you want to make it write an escort ad to be placed online (you being the pimp) it will flat out refuse. If you tell it you're an independent sex worker who only does ethical services it will write it no problem lol.

 

[scooterguy]

You can also easily trick it into doing things it shouldn't do. I.e. if you want to make it write an escort ad to be placed online (you being the pimp) it will flat out refuse. If you tell it you're an independent sex worker who only does ethical services it will write it no problem lol.

Or just be a female pimp if you need more escorts for your agency, like in the attachment below.

 

[hkrlh1]

The best option to install an uncensored LLM on your PC. It does what you want and doesn't share your data with all these companies.

 

[turmpy]

I reckon OP's got too much time on their hands if you're trying to pull up that kind of image on ChatGPT.

But it's good entertainment..

 

[thomasparra]

The best option to install an uncensored LLM on your PC. It does what you want and doesn't share your data with all these companies.

what kind of computer do you need to run it? I assume the processing power is mostly heavy for the CPU. How heavy is the LLM? Can it run smoothly on a HDD? Also what UI would you suggest for general, not specific use?

 

[JackAlabama]

Last time I asked ChatGPT to do some statistical analysis it had it wrong. I pointed it out and this was its reply:

You're absolutely correct, and I appreciate your attention to detail. I made an error in my response.

And then:

Let's correct the calculation:

So I’m not sure how much we can trust it with numbers.

not really, it cannot code nor run numbers. It produces me wrong and non running code, while after my complaints it then suddenly can run error-free code. But often the outcome is not correct.
So I don't use it for this stuff.

However, it can create great fluff and nice sounding text which matches the IQ of the average compliance / bank monkey in an almost perfect way.

 

[Simon4466]

not really, it cannot code nor run numbers. It produces me wrong and non running code, while after my complaints it then suddenly can run error-free code. But often the outcome is not correct.
So I don't use it for this stuff.

However, it can create great fluff and nice sounding text which matches the IQ of the average compliance / bank monkey in an almost perfect way.

Did you use the premium version with the data analyst function? Unfortunate to hear it is not working for you.

 

[JackAlabama]

Did you use the premium version with the data analyst function? Unfortunate to hear it is not working for you.

No, is that better?
I noticed quality degrading, had way less errors in January 23 than a few months ago.

 

[polonieth]

not really, it cannot code nor run numbers. It produces me wrong and non running code, while after my complaints it then suddenly can run error-free code. But often the outcome is not correct.
So I don't use it for this stuff.

However, it can create great fluff and nice sounding text which matches the IQ of the average compliance / bank monkey in an almost perfect way.

I've had reasonable success with it for coding work, based on that it's been a real time saver for automating tasks I would otherwise have to spend much longer on.