ChatGPT Conversations: Can we test if you are smarter than me?
A question that has come up with regard to AI and concerns in the field by researchers and the general population. “What if our intelligence is suddenly eclipsed by an on switch in a computer in a datacenter somewhere in the United States or China?” Publicly GPT-* are generative algorithms with the ability to infer from data it is trained on. ChatGPT with GPT-3.5 has something to say in its defense regarding limitations of its algorithm. I ran a series of questions and tests asking it about its ability. The soundness of its responses are not in question as it is inferring from static texts and that can be printed straight from Wikipedia.

Listening to News stories I have noticed misunderstanding about AI.and so do the developers at OpenAI. The first line is not created with inference other than the understanding that in t his situation regarding the ethics and comparisons to humans uit should make itself known for sake of argument. It says plainly that it is not a conscious machine. As in it does not have a soul and is not aware of you other than what has been typed into the prompt as the same with other Natural Language Models. The second paragraph it says it can perform various linguistic tasks
To follow the rest of the paragraphs are essentially an advertisement and simple context to compare between this algorithm and a human. Regardless, it is a very capable algorithm. Lets look at some other things to ask about itself and other NLP (Natural Language Processing) tasks regarding its usefulness as a tool in _AugmentedIntelligence in the future.

Here is another message from ChatGPT after being asked directly if it can infer from its infer or learn from its own inferences.

Not sure if you have heard of Cleverbot, “CB” was able to learn from what users were typing to it if i so happened to learn something from the user to be more human-like. As it sends we cannot force ChatGPT to learn from what we say or from what it says itself. So everything is a basic infer from the text data it was trained on. Text data means sources like Wikipedia, WikiData etc.

So I got a little more specific about what the algorithm can do. Here is an example of a direct question asking for heuristic for programming in C++.

Simple as that some tips for working with C++ code.

Some I have heard of and some I have not.

Some Philisopjical Insigt that it can provide

Not willing to be personal but is willing to provide insight on why some people would not want to exist.

A topic more controversial but could be down the road?




Filed under: Uncategorized - @ August 4, 2023 2:04 pm