Way off topic AI question
Thread Starter
|
Hotshot
Joined: Jan 2001
Posts: 16,882
Likes: 3,126
From: SE Wisc. (the Rust Belt)
Way off topic AI question
My question is for the more computer savvy among the FTE collective. What is Artificial Intelligence? Yeah, I know the basics, it's a computer system that can think for itself and can solve problems on it own but what is it? It seems like all of a sudden it popped up and now everyone one is using it, websites and control systems for almost everything including automobiles. If this wide spectrum of usage is being used where does it come from? Can you buy a premade software to install in applications? Did one source come up with it and then sell it to other vendors? How did start? If I wanted to use AI in an application do I buy a software package from some company? I assume once someone developed it it was copied by many others.
I'll probably never need to use AI for my personal use, I'm just curious. I know there are computer people here on FTE that can provide an answer.
I'll probably never need to use AI for my personal use, I'm just curious. I know there are computer people here on FTE that can provide an answer.
Parts Nerd Extraordinaire
Joined: Sep 2010
Posts: 4,367
Likes: 1,785
From: Western NY
It all comes down to training and pattern recognition. The youtube channel 3 Blue 1 Brown has a whole series for the intelligent layman. Since much of the hubbub is around ChatGPT and similar LLM (large language model), I'll point you there. As for adding "AI" to a program, it comes down to adding/training a model for the specific tasks. Specific engines and models have been built by large organizations. As for self installing, these large LLMs are too big and the engine will run too slow on commodity computers like we have in our homes, and so most of these models and engines are available on hosted infrastructure (think large data centers), hence most AI usage being subscription based to support the ongoing costs of running these computers.
If you wanted to see an AI chatbot using a LLM, this is a good first toe in the water experience: https://chatgptnologin.com/
Edit: My answers here are very high level and simplistic and not necessarily all encompassing.
If you wanted to see an AI chatbot using a LLM, this is a good first toe in the water experience: https://chatgptnologin.com/
Edit: My answers here are very high level and simplistic and not necessarily all encompassing.
Last edited by bmoran4; Apr 28, 2024 at 03:05 PM. Reason: Clarify
Fleet Owner
Joined: Dec 2000
Posts: 25,378
Likes: 5,387
From: Central PA
1. My son is a software engineer. He works for Merck Pharmacuetal. He says b moran is right on. He says 3 Blue, 1 Brown is a good channel.
2. He says Bob Jones should copy and paste that question to Google Gemini and see how AI answers that question.
Also, Zach says he doesn't think AI would work for an industrial application. Unless you need it for creative ideas.
2. He says Bob Jones should copy and paste that question to Google Gemini and see how AI answers that question.
Also, Zach says he doesn't think AI would work for an industrial application. Unless you need it for creative ideas.
Thread Starter
|
Hotshot
Joined: Jan 2001
Posts: 16,882
Likes: 3,126
From: SE Wisc. (the Rust Belt)
I got about 1/8 through that video before I faded away. Way too much for me to comprehend.
I was just wondering if all AI was generated from one source or are there many vendors who offer it. I am also curious as to how it was developed, did one person or team start working on and other were able to some how acquire the elementary form and further develop it further for their own specific use.
As for industrial use I am involved in industrial automation, test cells, assembly stations and robotics. There has to be a place for AI in these types of uses.
I was just wondering if all AI was generated from one source or are there many vendors who offer it. I am also curious as to how it was developed, did one person or team start working on and other were able to some how acquire the elementary form and further develop it further for their own specific use.
As for industrial use I am involved in industrial automation, test cells, assembly stations and robotics. There has to be a place for AI in these types of uses.
Parts Nerd Extraordinaire
Joined: Sep 2010
Posts: 4,367
Likes: 1,785
From: Western NY
There are numerous AI vendors, engines, and models. The start of neural networks, the main component driving most current AI technologies, goes back to the early 40's. As to the "first", it is highly dependent upon numerous qualifiers. A few names to be thrown into the hat:
1943 - Warren McCulloch & Walter Pitts developed the first models of neural networks (biological brain)
1957 - Frank Rosenblatt demonstrated the first trainable neural network, the Perceptron
1959 - Bernard Widrow & Marcian Hoff developed models named "ADALINE" and "MADALINE". ADALINE was developed to recognize binary patterns, MADALINE used an adaptive filter that eliminates echoes on phone lines. These are commonly recognized as the first neural network applied to a real world problem, and is still in commercial use in areas like air traffic control.
1943 - Warren McCulloch & Walter Pitts developed the first models of neural networks (biological brain)
1957 - Frank Rosenblatt demonstrated the first trainable neural network, the Perceptron
1959 - Bernard Widrow & Marcian Hoff developed models named "ADALINE" and "MADALINE". ADALINE was developed to recognize binary patterns, MADALINE used an adaptive filter that eliminates echoes on phone lines. These are commonly recognized as the first neural network applied to a real world problem, and is still in commercial use in areas like air traffic control.
Phantom of the Phorum
Joined: Jan 2011
Posts: 5,472
Likes: 1,696
From: Kentucky
There are numerous AI vendors, engines, and models. The start of neural networks, the main component driving most current AI technologies, goes back to the early 40's. As to the "first", it is highly dependent upon numerous qualifiers. A few names to be thrown into the hat:
1943 - Warren McCulloch & Walter Pitts developed the first models of neural networks (biological brain)
1957 - Frank Rosenblatt demonstrated the first trainable neural network, the Perceptron
1959 - Bernard Widrow & Marcian Hoff developed models named "ADALINE" and "MADALINE". ADALINE was developed to recognize binary patterns, MADALINE used an adaptive filter that eliminates echoes on phone lines. These are commonly recognized as the first neural network applied to a real world problem, and is still in commercial use in areas like air traffic control.
1943 - Warren McCulloch & Walter Pitts developed the first models of neural networks (biological brain)
1957 - Frank Rosenblatt demonstrated the first trainable neural network, the Perceptron
1959 - Bernard Widrow & Marcian Hoff developed models named "ADALINE" and "MADALINE". ADALINE was developed to recognize binary patterns, MADALINE used an adaptive filter that eliminates echoes on phone lines. These are commonly recognized as the first neural network applied to a real world problem, and is still in commercial use in areas like air traffic control.
(Come on, someone else was thinking it)
Trending Topics
old and in the way
Joined: Jul 1997
Posts: 5,921
Likes: 1,052
From: Lovely Hueytown Alabama
Bob,
Years ago I worked for a major steel producer. I was into large mainframe computers. In 1982 my boss sent me and another guy to an "AI" programming course. Basically you write a computer program (code... we used Cobol back then) and teach it the answers to questions. those questions and answers built the knowledge base for an AI system.
They wanted a system that could control a blast furnace during operation. They tried for several years but could never get it off the ground. Basically because there were too many variables in blast furnace operation and computer power/storage couldn't handle it efficiently... It was a cool project back then but It was too slow. Think about a situation that comes up and you want the program to sift through millions and millions of situations it has stored to find the answer and fix the situation.
Todays smart watches outshine the old mainframes we used. One company that did find a use and it was successful was Hallmark. Remember the Kiosks where you designed your own greeting card (from a limited number of choices)? Back then that was AI getting it's start.
Resistance is futile... You will be assimilated !
john
Years ago I worked for a major steel producer. I was into large mainframe computers. In 1982 my boss sent me and another guy to an "AI" programming course. Basically you write a computer program (code... we used Cobol back then) and teach it the answers to questions. those questions and answers built the knowledge base for an AI system.
They wanted a system that could control a blast furnace during operation. They tried for several years but could never get it off the ground. Basically because there were too many variables in blast furnace operation and computer power/storage couldn't handle it efficiently... It was a cool project back then but It was too slow. Think about a situation that comes up and you want the program to sift through millions and millions of situations it has stored to find the answer and fix the situation.
Todays smart watches outshine the old mainframes we used. One company that did find a use and it was successful was Hallmark. Remember the Kiosks where you designed your own greeting card (from a limited number of choices)? Back then that was AI getting it's start.
Resistance is futile... You will be assimilated !
john
FTE Stories
Ford Trucks for Ford Truck Enthusiasts
10 Things Every Truck Owner NEEDS (2026 Edition)
Michael S. Palmer
Rezvani's Latest Post-Apocalyptic Monster Is a Ford F-150 Raptor Underneath
Verdad Gallardo
Top 10 Most Expensive Ford Trucks Ever Sold on Bring a Trailer
Joe Kucinski
2027 Ford Super Duty Buyer's Guide (Every Model, Engine, & Package)
Brett Foote
Top 10 Ford Truck Tragedies
Joe Kucinski
AEV FXL Super Duty - the Super Duty Raptor Ford Doesn't Make
Brett Foote
Lobo Vs Lobo: Proof the F-150 Lobo Should Be Even Lower!
Michael S. Palmer
Ford's 2001 Explorer Sportsman Concept Looks For a New Home
Verdad Gallardo
10 Best Ford Truck Engines We Miss the Most!
Joe Kucinski
Thread Starter
|
Hotshot
Joined: Jan 2001
Posts: 16,882
Likes: 3,126
From: SE Wisc. (the Rust Belt)
John I've been in the industrial controls industry for over 30 years and I've seen many changes. When I started PLCs were just starting, the industrial heat company I worked for used relay logic for everything and finally they hired a PLC maestro, when he walked into a room to program a PLC everyone had to be quiet and not bother him. His laptop had a 4" screen and weighed about 10 pounds. He was, as he thought, worshipped. Now most low level maintenance guy can set up and program PLCs and other complicated equipment. Everything was 120 VAC and hard wired and run through metal conduits, now pretty much everything is 24 VDC or ethernet controlled and run through premade cable in wire trough. Safety circuit are a lot more complicated, before we'd use a simple ice cube relay to control a safety circuit now we have redundant safety relay with timing systems to make sure the circuit isn't defeated or compromised.
A lot has changed in my little world just like everywhere else. , .
A lot has changed in my little world just like everywhere else. , .
Thread
Thread Starter
Forum
Replies
Last Post
Delzell
Audio & Video Systems, Navigation, Satellite Radio & Mobile Electronics
1
Jun 3, 2007 02:31 PM