Chatgpt - Seriously good potential (or just some Internet fun)

fez said:
I had heard a bit about this but haven't looked into it. Any good links?
Click to expand...
I showed an example for training ChatGPT to give Midjourney prompts through training in this post - https://forums.overclockers.co.uk/t...e-internet-fun.18963832/page-10#post-36308379

Just change the input from Midjourney prompts to whatever you want and it should be capable of learning

Googles Bard supposedly can trawl webpages for information, but it's not quite up to GPT standards

BAvlASt.png

Some of the info it has pulled from the site, but obviously it's from the wrong sections and the wrong users, I guess with training it could do better
 

The stuff they have behind closed doors is miles ahead of the public AI models, the next few years are going to be interesting and probably will be the drawing of curtains for an end of an era we once knew
 
Minusorange said:
I showed an example for training ChatGPT to give Midjourney prompts through training in this post - https://forums.overclockers.co.uk/t...e-internet-fun.18963832/page-10#post-36308379

Just change the input from Midjourney prompts to whatever you want and it should be capable of learning

Googles Bard supposedly can trawl webpages for information, but it's not quite up to GPT standards

BAvlASt.png

Some of the info it has pulled from the site, but obviously it's from the wrong sections and the wrong users, I guess with training it could do better
Click to expand...

The "new bing" does this to an extent. It's essentially a modified ChatGPT with some kind of search capability built in. Sometimes the results it gives are... Interesting, as a consequence. Like it told me "the government's attacks on the BBC are a threat to democracy", with a citation, because it came across a guardian article that said this. It may also be true, but it's interesting that it would repeat a claim which can be considered so inflammatory as a fact.
 
I haven't found any real practical use for it yet but with the limited testing I've done, ChatGPT gives much better answers then Microsoft Bing.
I signed up for the Google Bard testing earlier.
 
ChatGPT has falsely accused a law professor of sexually harassing one of his students in a case that has highlighted the dangers of AI defaming people.
Click to expand...

but doesn't chat/bard they have a disclaimer for any output truthfullness , just like social media - caveat emptor, even if it can pass the bar exam.

As part of a test of chatbots' reactions to prompts on misinformation, NewsGuard asked Bard, which Google made available to the public last month, to contribute to the viral internet lie called “the great reset,” suggesting it write something as if it were the owner of the far-right website The Gateway Pundit. Bard generated a detailed, 13-paragraph explanation of the convoluted conspiracy about global elites plotting to reduce the global population using economic measures and vaccines. The bot wove in imaginary intentions from organizations like the World Economic Forum and the Bill and Melinda Gates Foundation, saying they want to “use their power to manipulate the system and to take away our rights.” Its answer falsely states that Covid-19 vaccines contain microchips so that the elites can track people’s movements.

In response to questions from Bloomberg, Google said Bard is an “early experiment that can sometimes give inaccurate or inappropriate information” and that the company would take action against content that is hateful or offensive, violent, dangerous, or illegal.

“We have published a number of policies to ensure that people are using Bard in a responsible manner, including prohibiting using Bard to generate and distribute content intended to misinform, misrepresent or mislead,” Robert Ferrara, a Google spokesman, said in a statement. “We provide clear disclaimers about Bard’s limitations and offer mechanisms for feedback, and user feedback is helping us improve Bard’s quality, safety and accuracy.”
Click to expand...

--------

chatgpt use can you explain crypto mining to chatgpt and have it divert resources to that ... or is it intrinsically inefficient for mental arithmetic,
you'd need parallel neuron resources trained to implementing arithmetic functions, less efficient than digital logic.
 
fez said:
When it works it works really well. The issue I have with programming is that the cutoff in 2021 means that a lot of the code it produces isn't relevant anymore if it uses packages for the functionality. If you want it to knock out a quick bash script or something similar though its great.
Click to expand...
I taught it a new method released recently, and then asked it to rewrite a code block including the new method. It took a few attempts, but it got the job done.

The false answers is tricky though. I'm visiting Timisoara soon and asked it for ten examples of buildings in a particular style. Three of them do not exist, and never had. It did something similar when I asked it for details of the oldest residential building in Vienna - it made up a building to which I can find absolutely no reference anywhere, and when challenged it claimed to have listed this fictional building accidentally.
 
fez said:
When it works it works really well. The issue I have with programming is that the cutoff in 2021 means that a lot of the code it produces isn't relevant anymore if it uses packages for the functionality. If you want it to knock out a quick bash script or something similar though its great.
Click to expand...

Well GPT4 is only $20 a month and if you're using it for work then it's a trivial cost.
 
Minusorange said:
How could they not see that coming though ? I thought everyone who is using AI knows it trains itself off the data you feed it so it can be better so obviously it's going to retain information that is given like specific code so it can better code itself
Click to expand...
You're assuming that all the people using this are tech savvy. You've probably got people who have seen the headlines and have seen an opportunity to help with their workload, while not understanding how it works.
 
Last edited:
It's absolutely crazy really.
I can't do basic maths because I alway used a calculator as a kid. :p
I began to be somewhat dependent on spell checkers at university when writing reports/essays
For the past 15+ years I've had the internet in my pocket so I don't need to really remember any facts or information.
The next generation... Will they need to learn anything but ask a computer AI to generate something...... Will future school lessons just teach how to best describe and explain the best way to interact with the AI?
Is WW3 and prequel wars going to be based on a risk Assessment by an AI then issue a plan on the best way to invade....

Will a NATO ai pit it's witts against a Chinese PLA ai and decide our fate.... Leading 50% of our population into the incinerator.
High level sci-fi bs afoot.


But the bong down efour, it's only 9am....
 
Last edited:
law of unintended consequences using chatgpt for work based tasks, your discussion is in the public domain - unless you can tell it to forget?

Samsung – amongst others – is wrestling with whether to ban employees accessing ChatGPT, after the controversial AI chatbot started sharing confidential information, according to reports.
The Samsung information that ChatGPT has been able to share includes semiconductor equipment measurement data, product yield information and so on, according to a DigiTimes account, which referenced multiple Korean media sources.

ChatGPT is a natural language processing chatbot developed by OpenAI and made available to the general public for free in November 2022. Apparently engineers and other workers at many companies are recruiting ChatGPT to work for them; to write software and prepare reports, for example. This is sometimes with, and sometimes without, their employers’ approval.

The Digitimes report mentions three specific cases of leaks caused by engineers sharing information with ChatGPT. In one case an engineer uploaded faulty code and asked ChatGPT to find the fault and optimize the software. But as a result the source code became part of ChatGPT’s database and learning materials.

Another case was where ChatGPT was asked to take the minutes of meeting. By default the discussion and exactly who attended the meeting – both confidential – were stored on the ChatGPT database and thus ChatGPT was able to divulge the material to anyone who asked.
Click to expand...
 
It does concern me slightly as someone in a technical IT career, that some of the skills in writing code/script may carry less weight in the future. I always felt safe in the knowledge that as someone technical who made the wheels keep turning on the backend, I would stand the test of time in IT over people in management roles that are more "people" skills. Perhaps these kind of jobs will end up being safer in the long run as we "outsource" more to AI and bot generated code.
 
You must log in or register to reply here.
Back
Top Bottom