ChatGPT and Bing AI might already be obsolete, according to new study

[Windows Central]

A new study reveals that scientists are on the verge of a new breakthrough with a new neural network technology. It has the capability to make generalizations about language, and scientists claim that it might be just as good as humans, especially when folding new words and applying them in different settings and contexts, ultimately presenting a life-like experience.

ChatGPT and Bing AI might already be obsolete, according to new study : Read more

 

[fjtorres5591]

"ChatGPT and Bing AI might already be obsolete..."​

Or they might not.

Go deep enough in the source and you find:

"Melanie Mitchell, a computer and cognitive scientist at the Santa Fe Institute in New Mexico, says this study is an interesting proof of principle, but it remains to be seen whether this training method can scale up to generalize across a much larger data set or even to images."

That is a polite way to say that there are a couple lightyears to cover from a small ivory tower test under controlled conditions and commercially useful production software.

Two points to remember:
1- neural networks are not a new technology and, in fact, LLMs are the output of existing neural networks. Neural networks have been all over doing productive work for decades. (Since the 50's_ in fsct.) The OP is simply discussing one of thousands(more?) out there.

2- Neural networks, LLMs, and other forms of "AI" (which is in no way intelligent, and lacks initiative or agency) is just software and more bluntly, software "plumbing". Anybody who has been around this territory long enough will remember similar "world-changing breakthroughs" and software plumbing "wars". The OOP wars, the UNIX wars (" my UNIX is more open than yours!"), RISC wars, OS wars. And now, "AI" wars.

Time to take a step back and remember a mantra from the 90's: people don't
buy technology--languages, OSes, architectures; plumbing in any form--they buy solutions.

LLM tech has its roots in "artificial intelligence" language processing concepts going back to Eliza in the 60's, with regular "world changing breakthoughs" every few years. LLM itself goes back nearly a decade and it is only now (barely) starting to be useful.

The whole world of "AI" is built on hype so take any breathless reports of "new! improved! with a pound of salt. Let's see it get out in the real world dealing with real people and real problems first.

 

[ShinyProton]

As if neither OpenAI or Microsoft would not adapt to changes in this field.