First AI actually was made in the 1950's as a nural network called a percepitron, an analog imaging computer. We still use it mathematically (matrices).
I ran a dial up BBS in the early 90s and I figured out how to launch an Eliza-like chat system instead o the default SysOp chat so it would talk to people when I wasn't around. More than once, I got home from school and watched someone arguing with it for several minutes and really not "getting" it was a bot.
A chat program called Dr Sbaitso came with my Sound Blaster ISA card. It was supposed to be a psychologist, but it said everything out loud from speech synthesis. It also had some hidden commands for goofy things. Like entering 40 column mode, or reading a text document. I got the Beowulf manuscript from Project Gutenberg and had Dr Sbaitso read it aloud. There are many references to "honeyed mead" within the text, and Sbaitso was hilarious in its attempt to say it, sounding like a discount Dalek, yelling "HONEY-ED ME-AIDE!" each time. Creepy outro BTW.
Fond memories of a variant of Eliza I had on the family PC in the early 90s. My younger brother was completely unaware that it logged all of the inputs to a text file. So much torment was inflicted, he had no idea how I knew so many embarrassing secrets... Another similar chatbot I used to enjoy was AGT, (Automatic Goblin Therapist) a World of Warcraft addon that could respond to whispers for you. I used to use it to ward off gold spammers and beggars, and post screenshots of the inevitability hilarious results to amuse guildmates.
I thought it wasn’t quite up to your usual standard, but was very surprised when I found out how it was written. Not sure wether to be impressed or worried.
I thought this episode seemed a bit rubbish until I reached the end, which earned a "like" from me. I also spotted emacs in doctor mode in the video, which reminded me that I have not tried it for ages, and only then did it a few times for the novelty value, because it is not really useful for anything. I have been using emacs nearly every day since the late 1980's, so visited my already open emacs window, and typed "ESC-x doctor", and yes, it still works! "Works" in this context comes with a lot of caveats! 🙂
Thanks - it was a risk to do this episode as some people will watch this as their first video on the channel and may just never come back - but wanted to show just how easy it is to exploit :)
Am I the only one who cannot bring himself to address a chatbot, home assistant, self-scan grocery terminal, etc, without unbridled contempt? I *despise* these things, especially their blatantly transparently contrived attempt to appear and "behave" human. To quote the great engineer, as I say every time I have to use one of these god forsaken things, "Up your shaft."
ChatGPT has two major flaws: its model uses static data, meaning it does not have access to the internet, for instance, to updates and it is *way* toned down. It tries so much to not offend anyone that sometimes hurts the interactivity and accuracy of the answers.
My favorite natural language processor is the Inform Virtual Machine in a well coded text adventure.... I remember trying out eliza when I was about 9... I was unimpressed. Still, I may have it on disk for CP/M... And, no, I didn't notice it, but was only halfway paying attention, doing post-session prep for my Alien RPG group...
In typical youtube fashion, I wasn't paying full attention (what with there being no Sinclair content!), but I don't think I heard you use the words 'Expert System' which is what we were taught at school in the 1980s as the answer to rules based problems. Did like the reference to choose your own adventure books though!
So I don't know if this works with every version of Eliza, but on the Acorn Archimedes school computers, if I typed "How do you do" in Eliza, it caused it to crash.
An interesting take, and thank you for the videos you make, enjoyable content well thought through and put together ^^. Though... now for the disagreement. Everything we all are comes from what already is. The is no original thought, there is only novel adaptation of existing knowledge and personal experience (combinations of this can be unique, such as to denote personality). Whilst its entirely possible to invent anything we want to, unless its connected to something which already exists, we have a word for this type of output, "Nonsense". The AI in ChatGPT is designed avoid generating nonsense (though it can capably lie, which is the border of nonsense). I suggest, try apply your criticism of current machine learning and GPT to yourself (as another entity which works by a set of learned and given rules), and realise you are entirely to be criticised or celebrated under the same criteria (e.g. best not to diminish the value in the ability to process a complex set of rules). I don't like the modern trend to make digital replacements for humans at the level of the social and the congnative, but I worst dislike the diminishing of machine learning and AI in a way that minimises us all (as beings which the AI is a reflection of); Any system capable of adaptive behaviour to the extent it can generate unique content should not be dismissed or diminished, because doing so diminishes and dismisses ourselves (e.g. enslaving intelligence is never a good thing IMO).
If RUclips is that bad, why do you watch content and comment on it? If RUclips is like some kind of "big brother" why are there millions of videos on RUclips and probably billions of comments, most of which go completely unmolested, even if they're controversial? And you can watch it for free, plus they even pay RUclipsrs for getting views. Yes, the barstewards at RUclips are pure evil. 😂
Please make sure you watch right to the end of this one, especially if it seems a little 'off' to you :)
Simply because of the subject I guessed you'd get ChatGPT to write the script. LOL.
First AI actually was made in the 1950's as a nural network called a percepitron, an analog imaging computer. We still use it mathematically (matrices).
I ran a dial up BBS in the early 90s and I figured out how to launch an Eliza-like chat system instead o the default SysOp chat so it would talk to people when I wasn't around. More than once, I got home from school and watched someone arguing with it for several minutes and really not "getting" it was a bot.
A chat program called Dr Sbaitso came with my Sound Blaster ISA card. It was supposed to be a psychologist, but it said everything out loud from speech synthesis. It also had some hidden commands for goofy things. Like entering 40 column mode, or reading a text document. I got the Beowulf manuscript from Project Gutenberg and had Dr Sbaitso read it aloud. There are many references to "honeyed mead" within the text, and Sbaitso was hilarious in its attempt to say it, sounding like a discount Dalek, yelling "HONEY-ED ME-AIDE!" each time.
Creepy outro BTW.
That Eliza/Parry convo is priceless.
lol love the bombshell at the end there. And no, I and most people wont notice.
Scary really!
Fond memories of a variant of Eliza I had on the family PC in the early 90s. My younger brother was completely unaware that it logged all of the inputs to a text file. So much torment was inflicted, he had no idea how I knew so many embarrassing secrets...
Another similar chatbot I used to enjoy was AGT, (Automatic Goblin Therapist) a World of Warcraft addon that could respond to whispers for you. I used to use it to ward off gold spammers and beggars, and post screenshots of the inevitability hilarious results to amuse guildmates.
I thought it wasn’t quite up to your usual standard, but was very surprised when I found out how it was written. Not sure wether to be impressed or worried.
I asked chatgpt if it knew Eliza. It said "Sure!" And gave me a history lesson on Eliza lol
Best episode yet! naughty, naughty.. :)
I thought this episode seemed a bit rubbish until I reached the end, which earned a "like" from me. I also spotted emacs in doctor mode in the video, which reminded me that I have not tried it for ages, and only then did it a few times for the novelty value, because it is not really useful for anything. I have been using emacs nearly every day since the late 1980's, so visited my already open emacs window, and typed "ESC-x doctor", and yes, it still works! "Works" in this context comes with a lot of caveats! 🙂
Thanks - it was a risk to do this episode as some people will watch this as their first video on the channel and may just never come back - but wanted to show just how easy it is to exploit :)
Am I the only one who cannot bring himself to address a chatbot, home assistant, self-scan grocery terminal, etc, without unbridled contempt? I *despise* these things, especially their blatantly transparently contrived attempt to appear and "behave" human. To quote the great engineer, as I say every time I have to use one of these god forsaken things, "Up your shaft."
ChatGPT has two major flaws: its model uses static data, meaning it does not have access to the internet, for instance, to updates and it is *way* toned down. It tries so much to not offend anyone that sometimes hurts the interactivity and accuracy of the answers.
Wouldn't want to offend anyone, i mean that could prove far too much for millennial, being the victim and all!
I have Eliza for the Amiga and the Speccy, (somewhere), although I can't remember what the Speccy version was called, it was'nt Eliza.
I don't know how, but I feel like there was no soul in this episode,
and yes this is a compliment 😁
Can recall on the Amiga Format (number 18?) we not only had Eliza but NIALL (non intelligent AMOS language learner). Both were brilliant and fun.
Strangely ChatGPT has the same quirks/features/ issues as Eliza even today.
Or why not "Dr. Sbaitso" that came with the Sound Blaster 2.0? That was funny
I loved Dr. Sbaitso! Just Eliza with a TTS engine taped on, but still.
@@JosiahGould Yep hehehehe
anybody remeber the evil version of eliza called Azile?
Have you tried asking ChatGPT for an impression of Vogon poetry?
I hadn't - but now I'm going to! :)
My favorite natural language processor is the Inform Virtual Machine in a well coded text adventure....
I remember trying out eliza when I was about 9... I was unimpressed. Still, I may have it on disk for CP/M...
And, no, I didn't notice it, but was only halfway paying attention, doing post-session prep for my Alien RPG group...
The thing I noticed was the lack of depth on the subject
That's kind of the point (I hope) - whilst it's clever and all, ChatGPT still has a way to go to produce good quality content :)
In typical youtube fashion, I wasn't paying full attention (what with there being no Sinclair content!), but I don't think I heard you use the words 'Expert System' which is what we were taught at school in the 1980s as the answer to rules based problems. Did like the reference to choose your own adventure books though!
So I don't know if this works with every version of Eliza, but on the Acorn Archimedes school computers, if I typed "How do you do" in Eliza, it caused it to crash.
An interesting take, and thank you for the videos you make, enjoyable content well thought through and put together ^^. Though... now for the disagreement.
Everything we all are comes from what already is. The is no original thought, there is only novel adaptation of existing knowledge and personal experience (combinations of this can be unique, such as to denote personality). Whilst its entirely possible to invent anything we want to, unless its connected to something which already exists, we have a word for this type of output, "Nonsense". The AI in ChatGPT is designed avoid generating nonsense (though it can capably lie, which is the border of nonsense).
I suggest, try apply your criticism of current machine learning and GPT to yourself (as another entity which works by a set of learned and given rules), and realise you are entirely to be criticised or celebrated under the same criteria (e.g. best not to diminish the value in the ability to process a complex set of rules).
I don't like the modern trend to make digital replacements for humans at the level of the social and the congnative, but I worst dislike the diminishing of machine learning and AI in a way that minimises us all (as beings which the AI is a reflection of); Any system capable of adaptive behaviour to the extent it can generate unique content should not be dismissed or diminished, because doing so diminishes and dismisses ourselves (e.g. enslaving intelligence is never a good thing IMO).
I remember Eliza ... not very helpful in any situation :).
Yes, cheese worries keep me awake as well.
I still have the disc :)
And 10 bucks says that screwtube deletes my last comment.
If RUclips is that bad, why do you watch content and comment on it? If RUclips is like some kind of "big brother" why are there millions of videos on RUclips and probably billions of comments, most of which go completely unmolested, even if they're controversial? And you can watch it for free, plus they even pay RUclipsrs for getting views. Yes, the barstewards at RUclips are pure evil. 😂