May 18th ChatGPT connected to internet

[imnotexist]

Some new features as its connected to the internet, how do u guys see this? I see it as terribly fucking dangerous. these features are taken from the twitter thread:

1. Summary of a newsChatGPT can now summarize any news in a seconds.Prompt → What were the main points from the last United Nations climate change report released in 2023? Summarize with a short paragraph and a list.

2. Detecting a trendAs it is connected to the internet, ChatGPT can become your best business partner.Prompt → Provide a short analysis of the latest e-commerce trends and consumer behavior patterns in the first quarter of 2023

5. Stock Market AnalystTake advantage of real-time data to have ChatGPT analyze company results.Prompt → Based on the latest financial results of the tech companies, how can their shares evolve in the coming weeks?

 

[salsa20]

it is very good

 

[imnotexist]

it is very good

As a hacker? lol Yes probably. I'm here learning. But the effects it has on society, do u still think its good? I mean they found out by accident that thru few prompts from a different language AI taught itself fluent i believe arabic or a dialect of one.

 

[marauda18]

Too much hype, which is normal with new technologies.

If you understand how a GPT works or LLM in general, you will know the limitations.

With that said, if we are talking about the negative effects on society, the main problem here is people using it, not the model itself, it's not alive(well, depends on your definition of alive) but I think you got it.We have seen what simple machine learning algorithms on social media can do to polarize society and create endless cycles of bullsh!t to watch based on your own behavior.

Yet, knowing how the model works,I'm very impressed how 3.5 and specially the 4.0 version performed in more complex natural language tasks.

 

[imnotexist]

Based on ur response, I'm assuming u work with AI in some capacity. Sorry for my ignorance, without safety restrictions, how uhh, how available are these models of AI? I'm also assuming there would be levels of restrictions that can be bypassed, and some impossible as proprietary info. I just don't know. Because with it now connected to the internet, the level of malware would be escalated even into a higher frame, say to a next dimension of 0day type of attacks with the internet at its fingertips, right? Excuse me for what probably seems stupid questions, im trying hard to learn more than the shit college teaches.

 

[marauda18]

Based on ur response, I'm assuming u work with AI in some capacity. Sorry for my ignorance, without safety restrictions, how uhh, how available are these models of AI? I'm also assuming there would be levels of restrictions that can be bypassed, and some impossible as proprietary info. I just don't know. Because with it now connected to the internet, the level of malware would be escalated even into a higher frame, say to a next dimension of 0day type of attacks with the internet at its fingertips, right? Excuse me for what probably seems stupid questions, im trying hard to learn more than the shit college teaches.

I literally just run your raw statement above in GPT4, here is the answer :

AI models like ChatGPT are freely available to anyone with an internet connection. Anyone can use these models to generate text or perform other tasks, and there are no restrictions preventing people from doing so. However, there are certainly ethical and safety considerations to take into account when using these models.

One major concern is the potential for misuse of these models. If someone were to use a model like ChatGPT to generate convincing but fake news articles or social media posts, it could have serious consequences for democracy and public trust. Additionally, there is a risk that these models could be used to support malicious activities, like phishing scams or social engineering attacks.

In terms of malware attacks, it is certainly possible that AI models like ChatGPT could be used to generate phishing emails or other types of malicious content. However, it is worth noting that these models are not inherently more dangerous than other types of software. Like any software system, they can be vulnerable to security vulnerabilities or exploitation if they are not designed and deployed securely.

Overall, the availability of AI models like ChatGPT is both a benefit and a risk: they have the potential to unlock new possibilities for human creativity and innovation, but they also pose serious ethical and security concerns. It is important for individuals and organizations to approach these models with caution, and to use them in responsible and ethical ways.

What irony.

Yes, I work with AI in some capacity, but I don't have the knowledge to answer this properly, it's more like ethical and philosophical what is going to happen in the near future.

 

[imnotexist]

LOLLL

ChatGPT never ceases to amaze. One thing of some comfort, I suppose, is I was able to bypass the safety SOMEWHAT, and write malicious code, but u still have to know how to deploy it if you expect anything out of it. So not just anyone can use the AI maliciously. Matter of fact, most don't even know how to proper ask it prompts, nevermind do nefarious activities. So those using it for such things would be doing it without it anyhow , ive just found it to make redundant coding easier. THX CHATGPT LUL

 

[marauda18]

LOLLL

ChatGPT never ceases to amaze. One thing of some comfort, I suppose, is I was able to bypass the safety SOMEWHAT, and write malicious code, but u still have to know how to deploy it if you expect anything out of it. So not just anyone can use the AI maliciously. Matter of fact, most don't even know how to proper ask it prompts, nevermind do nefarious activities. So those using it for such things would be doing it without it anyhow , ive just found it to make redundant coding easier. THX CHATGPT LUL

Yeah, I use my brain and these models to create more stuff, not the model alone. it's a tool and it's amazing.

There are agents being tested now, like AutoGPT, to 'work independently' which is promising but not the apocalyptic version people are afraid or hoping for. But I believe we are not far from that either.