ChatGPT / Artificial Intelligence

pete

pete

Brassica Oleracea
Staff member
Administrator
Just opening up the topic to the group, as our staff was discussing an apparent AI that posted here.

My real life job is, "Cybersecurity Analyst." That means I work with viruses and other fun malware and the havok they cause. Before that, I was a computer tech. That generally meant working with broken hardware or software. I have limited experience coding. I could probably fool someone into thinking that I have HTML or Javascript skills. I did have some BASIC coding chops ... about 40 years ago.

There's a maxim that says that ANY repetitive computer task can be programmed. When ChatGPT was publicly available, a few months back, I checked out ChatGPT to see if it really could either help me with some of my repetitive tasks. It has literally blown me away:

"Chat, I'm going to give you a bunch of URLs. I want you to strip 'http://' and 'https://' from each, then add this tiny bit of known-good code. Here's an example of what I'd like the output to look like: [example]."

"No problem. Here's your output."

I've also had some great success at having ChatGPT spit out some Python code to work with an API (application programming interface) for a website to work at checking malware samples automagically. The only thing that limits me is that I need to be able to describe properly what I want. That sometimes takes a few attempts.

ChatGPT and other AIs are not perfect. I've had a few instances where ChatGPT has spat out some code that just doesn't work and it's bad enough that I can see why it doesn't work. That can also be educational, though.

So, anyone else using ChatGPT or another AI? Anyone else try the new ChatGPT-fortified Bing search engine?
 
I've asked it a couple of tricky questions and gotten flat out wrong answers. It may be great at writing code, but I don't trust its answers to most everything else to be true. If this takes over the world as we've been told, humanity is doomed.
 
I have been keenly watching as the screenwriters, actors, and such are going on strike asking for protection from the AI industry (amongst other asks). Lots of my computer friends have been playing with the technology. I have so many projects going on here at Casa du Glassa (think chickens, gardens, maintaining a park I build, and oh yeah ... those four music ensembles I founded/run/play in). So I don't really think I will dabble with AI in the near future. The older I get, the more projects I have--didn't think it would work like that after I retired.
 
BarrySachs said:
I've asked it a couple of tricky questions and gotten flat out wrong answers. It may be great at writing code, but I don't trust its answers to most everything else to be true. If this takes over the world as we've been told, humanity is doomed.
Click to expand...
Humanity was pretty good at dooming itself before there was AI, so my worry level isn't much higher than it was before. :D

I heartily agree with you that AI is definitely not always right (I've started playing with the Bing version of ChatGPT and Google Bard, so I'll change to "AI"). With coding, it either works or it doesn't and I've had to have ChatGPT, specifically, re-generate responses. However, while I think some of the incorrect responses are beta software being beta software, some of the problems do seem to be with how I've asked the question.

Speaking of Gandalfe's comment, there was the story about some movie studio wanting to capture extras' likenesses and use them forever, as they can be manipulated through AI.
 
Problem is when you start hardcoding rules and patches for specific scenarios, it’s no longer “intelligent”, it’s just a rule following machine like any other computer.

I saw a great example of this when an AI was fed thousands of pictures of cats in order to teach it to identify a cat. It was then able to identify any cat … until you give it a picture of a stuffed cat. So then you have to tell it stuffed cats are also cats. Works great until you give it a cat statute to identify. Now you have to tell it statues can also be cats. You see where this is going.

A toddler can easily handle such a task. A so called artificial intelligence still cannot. So there is no intelligence, just error prone grammar parsing and blind regurgitation of bad information it found on millions of web pages.

This is not to say AI may never truly exist, only that it ain’t here yet. What we’re labeling AI today is far from intelligent and downright dangerous mostly because we apparently trust that’s it’s answers are correct.
 
BarrySachs said:
Problem is when you start hardcoding rules and patches for specific scenarios, it’s no longer “intelligent”, it’s just a rule following machine like any other computer.
Click to expand...
I've given this some thought. I don't necessarily think that a "rule following machine" can't be considered intelligent. I think you can argue that instincts are hardcoded rules. Looking at it from a different angle, if something causes you pain, you're not going to do that thing again. That's a rule that can be expressed as a simple if/then statement. Hungry? You eat. Etc.

A statue of a cat isn't a cat. A stuffed cat doll isn't a cat. They're representations of cats. I don't know what you need to know to make the logical leap from "thing" to "representation of thing". I don't know if that's an expression of intelligence, either.
 
You must log in or register to reply here.
Back
Top Bottom