Banjo Hangout Logo
Banjo Hangout Logo

Premier Sponsors

2175
Banjo Lovers Online


Nov 28, 2024 - 9:53:47 PM
28393 posts since 6/25/2005

I just had an experience that shows that AI (or at least Google’s web version) is definitely artificial, but not particularly intelligent. I asked for the population of Redding, CA, where I once lived, in 1970.

Google AI confidently pegged it at 77,640, which I knew was ludicrously high…..So, I refined my query and asked for the 1970 census population of Redding. That was 16,659–in the area of what I remembered. So Google’s AI lacked the intelligence to query the easily available census number. Why would I (or anyone) trust such a poorly programmed tool?

Nov 29, 2024 - 3:01:49 AM
like this

42532 posts since 3/5/2008

Didn't someone here just mention that AI is only as smart as the folks that programmed it..?

Nov 29, 2024 - 4:26:07 AM
likes this

194 posts since 6/22/2016

Functioning artificial intelligence is nothing more than a fantasy that will leave our collective cultural memory in utter shambles. Unless the actual goal of AI is to further degrade an already very low cultural common denominator, it is simply not trustworthy for any practical purpose.

Nov 29, 2024 - 6:04:06 AM
like this

4820 posts since 4/22/2018

That element of AI is almost a novelty, writing essays and answering trick questions is a far cry from where it can actually be useful . Where AI is become huge is in industries where it can be used to take over tasks in fields such as procurement, human resources, marketing, facilities etc. A good example would be in a large organisation like the one I work for (20,000 employees), every 2 years we all need a lap top refresh, where you put a request in for a new one, specify which one you need and where/when you want it. This can all be done where the only human involvement is the delivery guy who knocks on your door and hands it to you.

Nov 29, 2024 - 9:51:23 AM
likes this

1919 posts since 11/10/2022

AI is just a contextual search engine using neural networks to weight and sum the outputs. Its been around since Kohunen sp? in 1964 but computers and databases just got fast enough for meaningful results. Its not human intelligence but rather a simulation of how animal brains learn and forget. A few years back the buzzword became "deep learning". It is only electrical and far simpler than neuralchemistry.

What we used to call image processing became CGI and is now also AI.

What we used to call voice synthesis is now AI.

What we used to call expert systems(logical decision trees using computers) is now AI

What we used to call guidance systems, sampling theory, cruise control, washing machine spin cycles etc etc is now AI

AI is just a meaningless buzzword.

But what has been done is contextual search engines can not automate out stock brokers and other data analytics. They used to use "fuzzy logic"...but that math is now called AI as well. All ver laughable, but non specific and becoming more useful(and wasteful).

Dec 2, 2024 - 7:07:06 PM

donc

Canada

7563 posts since 2/9/2010

https://www.youtube.com/watch?v=1VqpQWJbB5w         Has anyone seen these convincing 'AI' photos of popular cars from the 1950's being produced in 2024 ? These reproductions of the 56 or 57 Chevrolets look quite real although the actual copying is pathetically out of scale. They seem to create a realistic photo and place it in a realistic auto show backdrop. For some reason a Canadian location will not always allow me to copy a photo from Youtube but try to Google : " 1957 Chevrolet reproduction"

Edited by - donc on 12/02/2024 19:13:55

Dec 6, 2024 - 8:17:36 AM

KCJones

USA

3355 posts since 8/30/2012

AI is just a hot buzzword Machine Learning. It is not new, at all. Advances in processing technology have allowed a lot of fast progress with Machine Learning in the last 10 years. ChatGPT, Gemini, Alexa, and all the others are "Large Language Models", which is Machine Learning applied to written language. LLMs are a rudimentary application of Machine Learning, doubly so if you're using free public LLMs rather than paid private LLMs. The vast majority of commercial applications for Machine Learning are unrelated to LLMs. Do not judge Machine Learning by your experience with an experimental public beta test of a free chatbot service.

There is a reason we're recommissioning nuclear power plants to energize data processing centers that function as Machine Learning systems, there's a reason that the DoD is one of the largest purchasers of nVidia boards, there's is a reason that we're currently having a trade war related to advanced computer chips and restricting the export of those chips to other countries. There's a reason the top traditional Engineering (civil, mechanical, chemical) firms now have entire Software Engineering groups on their teams. And those reasons have nothing to do with a free web-based chatbot.

Imagine being able to simulate a battle 10,000 times to determine likely outcomes and refine your approach before you attack. Imagine being able to decypher codes instantly. Imagine all the work it takes to develop contingency plans for every imaginable circumstance, and being able to do that with the click of a button rather than an entire floor of intelligence analysts. Imagine a swarm of 5,000 autonomous micro-drones all synced together, dropped from a mothership flying at 60,000 feet, each with a 10 gram C4 charge attached to them, and the ability to automatically identify and seek enemy combatants, released over a battlefield.

Sufficiently advanced technology is indistinguishable from magic. Underestimate it at your own risk.

Edited by - KCJones on 12/06/2024 08:27:43

Terms of Use | Privacy Policy | Privacy Consent
Copyright 2025 Banjo Hangout. All Rights Reserved.





Hangout Network Help

View All Topics  |  View Categories

0.1416016