191 replies to this topic

[Stains_not_here_man]

You know how I feel about inflamatory language, but ironically they pulled it because it was making shit up - exactly as we have been saying throughout the thread.

Mando will tell you that since humans can also make things up that means it's an AI

[positiveContact]

Mando will tell you that since humans can also make things up that means it's an AI

Lol! No, I'll tell you that's not a reason to say something isn't AI. This is actually a failure of logic on your part.

You can't say something isn't AI because of "x" if humans also do "x" unless of course you don't think we are intelligent either.

[Stains_not_here_man]

Lol! No, I'll tell you that's not a reason to say something isn't AI. This is actually a failure of logic on your part.

You can't say something isn't AI because of "x" if humans also do "x" unless of course you don't think we are intelligent either.

I never said it "wasn't AI" because it makes things up. The point is that it isn't a "general" AI which has a specific meaning. It's a language model. That's a very specific type of AI and people are treating it like it's a different, more general kind.

[Stains_not_here_man]

Here, I'll let ChatGPT explain it to you.

Please explain why ChatGPT is not Artificial General intelligence



As an AI language model, ChatGPT is designed to perform specific tasks such as answering questions, generating text, and completing sentences. While ChatGPT can generate highly complex and coherent responses, it is not an artificial general intelligence (AGI) because it does not have a broad range of cognitive abilities and cannot perform a wide variety of tasks in the same way that a human can.

AGI refers to AI systems that possess a broad range of cognitive abilities that can be applied to various tasks and domains. These systems can learn and reason about different concepts, make decisions, and generalize knowledge to new situations.

ChatGPT, on the other hand, is a narrow AI system designed to perform specific tasks based on a pre-defined dataset and algorithms. It lacks the ability to reason abstractly, understand emotions, have consciousness, and perform tasks that require physical interaction with the environment.

While ChatGPT can generate highly realistic responses, it does not possess the cognitive flexibility and adaptability of an AGI. As such, it cannot be considered a general intelligence, but rather a highly sophisticated tool for generating text and answering questions within a narrow scope of knowledge.

This is pretty much what I've been saying all along. It can't reason, make inferences, apply general knowledge to new unfamiliar situations.

[positiveContact]

Ironic.

[Stains_not_here_man]

The funny thing is that I could probably get it to tell you why it IS an AGI too.

Because it doesn't actually "know" if it is or not. It just generates sentences one way or the other.

[positiveContact]

The funny thing is that I could probably get it to tell you why it IS an AGI too.

Because it doesn't actually "know" if it is or not. It just generates sentences one way or the other.

You have to look at this from chatgpt goals and training. Truth, accuracy, consistency are not necessarily the goals.

[Stains_not_here_man]

You have to look at this from chatgpt goals and training. Truth, accuracy, consistency are not necessarily the goals.

Oh, I know. It says right there on every chat that you have with ChatGPT that their goal is to make AI more natural and "safer" to interact with.

Which only furthers my point that people shouldn't be saying things like "ChatGPT wants..." or "ChatGPT thinks..."

[positiveContact]

Oh, I know. It says right there on every chat that you have with ChatGPT that their goal is to make AI more natural and "safer" to interact with.

Which only furthers my point that people shouldn't be saying things like "ChatGPT wants..." or "ChatGPT thinks..."

I don't think this is always with the intent of saying it's actual intelligence. I say things like this about my own algorithms that are definitely not anything like AI. Those click bait headedlines are obviously something else.

Edited by positiveContact, 24 March 2023 - 07:03 AM.

[positiveContact]

Wiki article on AI Safety.  https://en.wikipedia.../wiki/AI_safety

 

If you assume this definition the word safety makes sense:

 

AI safety is an interdisciplinary field concerned with preventing accidents, misuse, or other harmful consequences that could result from artificial intelligence (AI) systems. It encompasses machine ethics and AI alignment, which aim to make AI systems moral and beneficial, and AI safety encompasses technical problems including monitoring systems for risks and making them highly reliable. Beyond AI research, it involves developing norms and policies that promote safety.

[Vagus]

break your computers before they brake you

[positiveContact]

https://arstechnica....with-the-world/

 

ChatGPT gets “eyes and ears” with plugins that can interface AI with the world Plugins allow ChatGPT to book a flight, order food, send email, execute code (and more).