OT: Open AI’s new model is insane

[NJPSU]

I'm far from being an expert on anything to do with computers. So, here's my understanding of AI.

AI is a machine. It doesn't think, it doesn't have feelings, it isn't sentient. All of those are attributes that some people give it.

What AI is is a really good search engine. It looks at indexes, mainly word indexes. I've heard it said that at any given time, Google has access to about 3% of the entire internet. Seems that the amount of data is limited. AI looks at those word associations that are input from the question asked. It finds other words associated with it on those indexes. Its all pattern recognition. When they find a consensus of identical associations, they use that for its output. That consensus is about 70%, which leaves a 30% room for error. To get any higher consensus would consume a lot more power.

Being a machine, it runs on an algorithm. Someone has to program it. If it can write code, its been programmed to do that. The Google search engine is programmed to show certain output. AI can also be programmed to do the same. Human bias can creep in pretty quickly.

The other thing to be aware of is the way data is indexed. That is done by humans, and, from my understanding, some of it is pretty bad.

I'm okay with AI doing some of the menial work. But, I'm not okay with AI doing important work. I don't want it setting climate or environmental policy, for example. Those things have way too many variables to be considered. It may be a good tool for finding some of the variables, but it’s only as good as the data going in, and the underlying programming. Things may get better as time goes along, but we're not there yet.

How do scientists make new discoveries? They experiment, analyze the results, adjust, experiment again, repeat until they get the desired result. AI can do this but infinitely faster than a human. That’s why the 60 minutes segment on AI talked about it creating new deadly chemical weapons.

To say AI is just going to be a good search engine is drastically underselling what it’s going to be capable of doing.

Many in the field feel we are already close to reaching AGI (Artificial General Intelligence) which matches or exceeds human intelligence across a wide range of cognitive tasks.

 

[91Joe95]

So what's AI? It's freaking math and statistics. That's it, it's a bunch of algorithms. I don't understand why people put it on a pedestal, or fear it. The people designing and running it are the problem. It will have its uses, but it also comes at some pretty significant economic and energy costs. Sure, it can greatly affect the health care industry and devalue doctors, but it could easily eliminate millions of government workers.

 

[91Joe95]

How do scientists make new discoveries? They experiment, analyze the results, adjust, experiment again, repeat until they get the desired result. AI can do this but infinitely faster than a human. That’s why the 60 minutes segment on AI talked about it creating new deadly chemical weapons.

To say AI is just going to be a good search engine is drastically underselling what it’s going to be capable of doing.

Many in the field feel we are already close to reaching AGI (Artificial General Intelligence) which matches or exceeds human intelligence across a wide range of cognitive tasks.

This is really nothing new - they've been using math models in nuclear weapons development and storage for decades. It's why they don't detonate nuclear weapons any more for tests.