For people who didn’t exist in 1985, the reason Paul Schindler is always complaining about copy protection is because before hard drives, software would be locked to the original diskette. When hard drives became more common, copy protection just meant you had to use the original floppy to run the program, and you couldn’t install it to the hard drive.
I remember watching this episode back in 1985 and I wondered how much things would change in 10 to 20 to 30 years. Well I have found out how much we have progressed with Artificial Intelligence over those periods of time. Basically Artificial Intelligence (or AI) has made our computer systems faster and more efficient.
25 years later and we finally have AI that can do what this gentleman pretended his system could. Which is getting a machine to do complex tasks through natural language, like structuring database queries and so fort with chatGPT. We were "nearly there" for 20+ years, until it finally clicked and the AI suddenly could do much more than anyone would dared dream of. We are certainly alive in one of the most interesting times.
the best thing, is, i think, that neural networks were already known, though they usually ran on perceptrons, which were way simpler than our current ones.
were those true for all the versions? unix for the ibm pc, i doubt most models would have that type of memory protection, given that that only began with the 286 (very slightly) and 386? am i wrong? and isn't the RAM limit based on the processor too? people already knew VAX hardware could do those things, after all.
So NLP and neural networks were known by computer scientists in the 80s but these are business people not scientists. They are building decision trees. This is not AI or ML. Computers were not fast enough and there was not enough data to train neural networks to do anything useful at this time.
There's actually a Computer Chronicles episode dedicated to neural networks. It's about as well as could be expected with such limited hardware. They even built some analog neural networks on integrated circuits. I think it gets mentioned in the neural net video.
Gary kildall was so far ahead of the time. He talked about these companies misusing the word AI like candy to sell their garbage products 40 years ago. I honestly think Gary was the FIRST person to ever recognize and point out the word AI being misused
Lmao and especially considering modern AI or voice assistants like Siri and databases in general are MASSIVE, Yet here this guy around 10 min in or whatever is telling Stewart and kildall that his software is so “smart” and “intelligent”, on a relatively EXTREMELY weak cpu (compared to today), not to mention low low ram and disk size, and processing power. No way could he have brought a piece of software that’s intelligent
Amazing how they were talking about grandiose AI computing on systems that are so comparatively weak today. Not even talking about those 8088/286 desktop systems but even the Cray 2 supercomputer that came out in that year is eclipsed 5-10x in performance by even the cheapest $20 disposable smartphone today.
NEO O there’s always going to be people with a solution in search of a problem. Speech recognition in 1985 was the equivalent of Amazon Echo, Google Assistant, and Apple Siri in 2019.
@@qsw4kchips827 It's public knowledge that they used TERCOM. en.wikipedia.org/wiki/TERCOM#Missiles_that_employ_TERCOM_navigation Besides, I worked with a guy who had worked on TERCOM.
Steven Burrell plus there wasn’t much pure research into AI that was even possible at the time. Rather than draining funding from practical AI research, they would have drained funding from impractical research and into practical defense projects.
When we first started opening pandoras box. AI absolutely exists now....unfortunately. At least right now our AI is severely limited. However, if we allow AI and quantum computers to work together...not good.
Not really. Run a spellcheck in Word for example, "too" and "to" are both correct words however it can work out from the context of the sentence whether the one you used is correct or not.
Watch @2:41 for the explanation. Gary hit it on the nail that the industry needs to determine what works and what sounds nice but doesn't really work. AI made huge strides since then but only in specific things, most largely invisible to the end user.
This episode brought to you by: 'If then else'; 'if then else', all about decision making.
10:35 Watching this while prompting ChatGPT to write a script to use for at work 😱
Then we can imagine, what would look like CHAT gpt, in the same frametime, if we think exponentially. As we push the limits of AI.
For people who didn’t exist in 1985, the reason Paul Schindler is always complaining about copy protection is because before hard drives, software would be locked to the original diskette. When hard drives became more common, copy protection just meant you had to use the original floppy to run the program, and you couldn’t install it to the hard drive.
does this old AI impress you?
Just as today, copy protection only annoys the people who legally bought the software. All the pirates remove the copy protection anyway.
I remember watching this episode back in 1985 and I wondered how much things would change in 10 to 20 to 30 years. Well I have found out how much we have progressed with Artificial Intelligence over those periods of time. Basically Artificial Intelligence (or AI) has made our computer systems faster and more efficient.
The level of simplicity in how Wendy explains the complexity of ai in this video is awesome. Most people don't speak like this anymore. 😍🤩😍
20:00 still relevant, interacting with a LLM in 2024 feels very much like a cooperative effort and experience
10:24 - This software is totally just "20 questions", except the computerised 20 questions hadn't been invented yet.
Dude was trying to call it AI software until Gary called him out on it, then he started calling it "Expert System"
Holy chit Gary kildall was WAYYY ahead of his time. Predicted the whole AI bs being misused as a word to sell Better 40 years ago
25 years later and we finally have AI that can do what this gentleman pretended his system could. Which is getting a machine to do complex tasks through natural language, like structuring database queries and so fort with chatGPT. We were "nearly there" for 20+ years, until it finally clicked and the AI suddenly could do much more than anyone would dared dream of. We are certainly alive in one of the most interesting times.
recheck the math
They are basically just talking about a bunch of simple control statements stacked together. Neural computers of today would blow their minds.
the best thing, is, i think, that neural networks were already known, though they usually ran on perceptrons, which were way simpler than our current ones.
were those true for all the versions? unix for the ibm pc, i doubt most models would have that type of memory protection, given that that only began with the 286 (very slightly) and 386? am i wrong? and isn't the RAM limit based on the processor too? people already knew VAX hardware could do those things, after all.
Juan Nunez the autofocus in our phone cameras is more powerful AI than anything on this show.
Do you have any example of AI in 2020 ?
What are "neural networks" actually doing ? (Fast parsing through a database is not intelligence.)
@@melmaciandissenter2324just 3 years. ChatGPT said hello
So NLP and neural networks were known by computer scientists in the 80s but these are business people not scientists. They are building decision trees. This is not AI or ML. Computers were not fast enough and there was not enough data to train neural networks to do anything useful at this time.
There's actually a Computer Chronicles episode dedicated to neural networks. It's about as well as could be expected with such limited hardware. They even built some analog neural networks on integrated circuits. I think it gets mentioned in the neural net video.
Trying to sell simple pre-defined decision trees as AI... 😅
It was almost 40 years ago :) May be 40 years later someone will say that ChatGPT had no relation to AI... We always want more and more from AI.
Until Gary called him out on it, then started calling it an "Expert System"
With CHAT-GPT here, it's quite intersting to see the progress
Those Lotus guys were full of shit. Trying to pass off their basic database as having ‘AI’… trying to capitalize on the latest buzz words.
Gary kildall was so far ahead of the time. He talked about these companies misusing the word AI like candy to sell their garbage products 40 years ago. I honestly think Gary was the FIRST person to ever recognize and point out the word AI being misused
Lmao and especially considering modern AI or voice assistants like Siri and databases in general are MASSIVE, Yet here this guy around 10 min in or whatever is telling Stewart and kildall that his software is so “smart” and “intelligent”, on a relatively EXTREMELY weak cpu (compared to today), not to mention low low ram and disk size, and processing power. No way could he have brought a piece of software that’s intelligent
@@davidt8087 My microwave oven has more processing power than those desktops!
Cognitivism is a very important science topic. Artificial Intelligence is possible.
Amazing how they were talking about grandiose AI computing on systems that are so comparatively weak today. Not even talking about those 8088/286 desktop systems but even the Cray 2 supercomputer that came out in that year is eclipsed 5-10x in performance by even the cheapest $20 disposable smartphone today.
And our most amazing tech today will be eclipsed in the future by fairly mundane devices. It's nice to appreciate things within their context.
The answer to almost everything in this episode is...Google.
+ ChatGPT
25:35 why were they so into speech recognition, its an annoyance at best
NEO O there’s always going to be people with a solution in search of a problem. Speech recognition in 1985 was the equivalent of Amazon Echo, Google Assistant, and Apple Siri in 2019.
Mára egész jól elbeszélgetek velük, már a saját nyelvemen is. :D
Probably 2018 soon . . .
3:39 That's what early cruise missiles used to fly to their targets.
How do you know
@@qsw4kchips827 It's public knowledge that they used TERCOM. en.wikipedia.org/wiki/TERCOM#Missiles_that_employ_TERCOM_navigation
Besides, I worked with a guy who had worked on TERCOM.
That's one of the methods. Another method was Dead Reckoning.
@@knerduno5942that was _before_ TERCOM *and Inertial Navigation.*
It feels so short sighted not to work for the defence department. They made most of our modern communication technology.
Steven Burrell plus there wasn’t much pure research into AI that was even possible at the time. Rather than draining funding from practical AI research, they would have drained funding from impractical research and into practical defense projects.
Gary Killdal taught at the Naval Postgraduate School in the engineering dept and worked on research projects there also.
7:31 : nice way of sugar coating the fact that you failed to develop natural language processing and resorted to selection based processing. 😂😂
When we first started opening pandoras box. AI absolutely exists now....unfortunately. At least right now our AI is severely limited. However, if we allow AI and quantum computers to work together...not good.
Seems like we made zero progress with AI and home computers..where is it?!?! LOL
+Neo Racer The effort is probably not worth the reward financially.
Not really. Run a spellcheck in Word for example, "too" and "to" are both correct words however it can work out from the context of the sentence whether the one you used is correct or not.
Watch @2:41 for the explanation. Gary hit it on the nail that the industry needs to determine what works and what sounds nice but doesn't really work. AI made huge strides since then but only in specific things, most largely invisible to the end user.
how are you doing
this aged well
Hard-coded if-then-else sold as AI. It's so cringy I am having a hard time watching this.
17:35 shows a more sophisticated interface based on sentence analysis