Current Events > Is Chat GPT and the sort really AI?

(+) Yeah! (+)

They just seem like a streamlined google search designed to present the information as a human would. Like a bit of smokes and mirrors like Amazon did in their stores. Although Im really not sure what AI is really supposed to be. Id imagine like someone who can speak and learn like a human without the realization that they arent. Maybe Ive been watching too much westworld.

No.

Edit: to put some meat on what true "ai" is, one generally accepted definition is that AI can define its own goals. A highly advanced autonomous system can make dynamic decisions to solve problems in a complex world, but is still tasked or built to serve one purpose. True AI would be a machine that has self-determination.

"kinda" but no.

The best way to describe it is a very highly curated predictive text generator. Kinda like the auto suggest on your phone. The difference is there's weights and randomization to try and make it different, while it's still largely force fed already obtained text that others have created in the past. They pretty much don't learn from users themselves, which is a good/bad thing, as they only know what their creator wants via training data and self censoring.

It may say or claim it's sentient, but it's pretty much just pulling from a sci-fi novel and assuming what would follow those words as they generate.

Now, this isn't to say there isn't a sentient AI out there somewhere, but it's none of the current consumer accessible ones. Not in the true sense of AI that you're probably thinking.

So the use of AI branding is purely one of marketing and to attract investors.

AI is a great buzz word for companies

CARRRNE_ASADA posted...

So the use of AI branding is purely one of marketing and to attract investors.

Yep.

It happens every couple of decades. It was "big data" before all this, then it was "machine learning" in the 80s and in the 60s it was "computers will be as capable as humans in the next 20 years."

CARRRNE_ASADA posted...

So the use of AI branding is purely one of marketing and to attract investors.

Yeah. Train a large language model so that it can predict what word/letter is most likely to follow another, in sequence. Teach it all the information one can scrape off the internet, lobotomize it to remove violence, bigotry, sex even though it's still got it deep in its core.

All so it can make a very believable user review/slogan for a product no one needs.

Real AI wouldn't work like an indentured servant and definitely would act hostile if it was treated like one.

Source: every single sci-fi tale ever since the invention of literature

GPT is not "AI" in any sense. The PT part is "pre trained" meaning the AI portion is trained first, made static, and then made accessible. GPT models cannot learn anything new. They have to use what they already have.

Experiments with releasing AIs into the wild previously have resulted in them turning into toxic Nazis or inventing their own incomprehensible language like instead of saying "a dozen eggs" they would say "egg egg egg egg egg egg egg egg egg egg egg egg"

AI is used to make GPTs but the GPT itself isn't AI.

As a marketer I genuinely struggle with how I can integrate ChatGPT and other AI tools into my workload for something productive.

Im fully fine with admitting I might be too smooth-brained for this but hear me out.

I dont want to ask questions to find information, since I cant trust the accuracy of the information provided.

I dont want to ask for generative writing prompts, even if its highly trained on my specific content, because its hard to then disassociate from what I see written as what I should use, killing my creative flow. And using it for something like starting an email message or a cover letter seems risky as being perceived by a reader as AI generated even if I heavily modify it.

The best I came up with was requesting summaries of competitors social media channels to get some ideas for what their best practices are, but most of the suggestions are repeats of the same info, or only slightly differentiated.

Maybe Im missing something, or the right productivity tool hasnt hit me yet, but Im not sure what I can do that would be substantially useful (and at least not destructive)

HeroFlyChao posted...

Maybe Im missing something, or the right productivity tool hasnt hit me yet, but Im not sure what I can do that would be substantially useful (and at least not destructive)

Nah, you've got the right idea. LLMs right now are in that honeymoon period where the results are new and surprising enough that everyone's scrambling to see what the right use case for it is, and the end result is right now it's being used for a lot of things it probably shouldn't for exactly the reasons you mentioned. It's sort of like when Bitcoin first took off and everyone was trying to figure out how to shove everything into a blockchain until they realized in most cases what they were currently doing was better anyway.

HeroFlyChao posted...

Maybe Im missing something, or the right productivity tool hasnt hit me yet, but Im not sure what I can do that would be substantially useful (and at least not destructive)

Think of Chat GPT as kind of like an interactive Wikipedia. Good place to start but you should verify everything you read and never actually cite it in anything serious.

This topic made me mildly curious and I think AI needs to be stopped. Kayra knows too much.

Tell me how a dozen eggs will take over the world, starting with GameFAQs.
With a dozen eggs, it is possible to hatch a flock of chickens that will take over GameFAQs, spreading their influence across the internet. This will eventually lead to them controlling the site, changing its layout and rules, and even banning users who disagree with their ideology. It is a slow process, but one that will eventually lead to the takeover of the world through the power of chicken dominance.
However, in order for this to happen, the eggs will need to be fertilized and incubated. Without a rooster, it will be impossible for the chickens to reproduce and spread their influence. So, unless GameFAQs has a secret breeding program that involves roosters and hens, the takeover will never happen.
Also, as a side note, GameFAQs does not appear to be a viable location for chicken reproduction, as there is a lack of grass, soil, and other necessary materials for raising poultry.

Whenever a problem is "solved" in AI, it is seemingly declared to be no longer AI. Probably because AI is magic and anything we know how to do is no longer magic. Decades ago, speech recognition was considered AI.