@@gavinathling I suppose that makes sense for a company called America Online and not Europe Online lol. I had quite a few international friends back then and we all used MSN Messenger to keep in touch.
What a deeply researched video. Besides all the stuff you talked about the evolution of chatbots in the video I’m also mind blown by the execution and delivery of this video. You guys are simply outstanding 💯
I don't know why this illusion surprises everyone so much. We grow up naturally personifying everything. Show someone an object, give it a name and short story, then destroy it in front of them. They'll be sad. We want to empathize and connect with everything.
@@ArjunBhattarai One good thing about being old (I'm not _that_ old by comparison) is that you're not dead yet. For now, I'll accept being called old; the alternative will come soon enough.
This is a really thought-provoking framing of chatbots. There's so much discussion about the "information" impact (e.g., replacing knowledge workers, generating misinformation, etc.) that I'd never thought about how strong of an emotional impact they can have.
😂👍 after the AI insanity Vergecast I had an itch to go looking for a summary vid for the history of AI development milestones to remind me of those I’ve lived through & fill in any gaps & highlight the lesser known in the zeitgeist/ media highlighted players. Or a James Burke’s Pinball Effect / Connections type vid, that would be cool. The first vid of a acceptable length at that moment that had some views claimed that computers were using tape drives in the ‘90s & so was either written by an AI or a 12yr old. So I quickly moved on and had other things to do. Lo & behold a week later the Verge itself posts exactly what I was looking for from a known credible source that I could rely on being well researched & refer to academic papers !!! Thx 🙏
What I learned from hanging with boys is that when their GF is talking just repeat the last word in their sentence and nod, and they think that they're already making a conversation. So basically this is how basic language models were developed.
Damn. You guys should see what I've put together over the last few days.. Got duelplex dialogue working as of tonight So you and bot can speak at the same time. TTS powered by Eleven Labs and technically no latency.
All I need is the Emacs doctor, and I'm planning to bring up in my next session how incredibly disappointed I was that this video made no mention of it. M-x doctor for life.
The description of transformers as “being able to see the entire sentence” is true about the type of model used in “Attention is all you need”, however the type of model used for ChatGPT and other similar chatbots *is* only predicting the next word (or token).
It's totally not because chatGPT reads like a human that I forgive its mistakes. It's because I know that despite it sounding so authoritative and confident, it doesn't always get the data right.
8:05 Sorry but the explanation about transformers is misleading, not simplified. Depende models were a think way before transformers, so saying AI before couldn't see more than 1 word in the text is not simplifying, but just wrong.
Chatgpt got 9 divided by 3 wrong. It seems awful at basic math. After I called it out it apologised and said I was right and that while it strives to get it right it doesn't always. Seriously.
I've tried a couple of these bots. And I can't really find any reason to talk to them. They don't really feel like anything that interests me. I usually just end up insulting them in an attempt to get a rise out of them. And move on. Nothing to see here...
@@paulu_ yeah that’s the only use I can think of. But I hate when companies use bots for customer service. They’re just there to annoy you until you find a human.
@djp1234 it helps you find information and parse through the parts you didn't ask about. Think of it as an enhanced search engine. It's a huge productivity booster for administrative office positions, programming positions... just those alone will revolutionize society. Programmers already have done that without the autocoding chatbots so just imagine
So raise of hands, who remembers SmarterChild?
AOL wasn’t big in Europe, but I’ve honestly never heard of smarter child.
Never heard of it before. First chatbot I used was Cleverbot
Sorry not to see RACTER mentioned here. That program was...bonkers...
yeah i used to basically red team smarterchild with my friends when i was 10. so many good nights berating it. i feel bad now.
@@gavinathling I suppose that makes sense for a company called America Online and not Europe Online lol. I had quite a few international friends back then and we all used MSN Messenger to keep in touch.
Where is Clippy? 🙃
Nostalgia 🥹
JusticeForClippy
lol did clippy actually ever help u? 😂
They killed Clippy like Kenny
@@gorgonzolariggs7710 he was my emotional support 🤣
What a deeply researched video. Besides all the stuff you talked about the evolution of chatbots in the video I’m also mind blown by the execution and delivery of this video. You guys are simply outstanding 💯
not me getting youtube ads of the EXACT SAME VIDEO
I don't know why this illusion surprises everyone so much. We grow up naturally personifying everything. Show someone an object, give it a name and short story, then destroy it in front of them. They'll be sad. We want to empathize and connect with everything.
"I know. I know. 1966 is a very long time ago."
Thanks for making me feel old.
Ditto. But it was :)
make you feel old? You are old. No one needs to remind you.
@@ArjunBhattarai One good thing about being old (I'm not _that_ old by comparison) is that you're not dead yet. For now, I'll accept being called old; the alternative will come soon enough.
@@ArjunBhattarai You're not old until you smell old.
@@ArjunBhattarai >no one needs to remind you<
but sometimes it's nice to forget that you're past your prime
It's cool seeing David Pierce back
True. He use to be my company at lunch on the times of 'Top Shelf '
This is a really thought-provoking framing of chatbots. There's so much discussion about the "information" impact (e.g., replacing knowledge workers, generating misinformation, etc.) that I'd never thought about how strong of an emotional impact they can have.
Brilliantly written and narrated!
"God help you part "really got me
I found the presented by SAP logo extremely distracting since it kept moving in and out of the video.
Just found myself in the very odd situation of having to watch a RUclips ad promoting this video before watching this video. Meta
4 minutes in and just watched a mid role ad of the video I am in the middle of watching…
7 mins in and once again we interrupt this broadcast to tell you you should watch this broadcast.
Clicked to see if Dr. Sbaitso was mentioned. Well done.
😂👍 after the AI insanity Vergecast I had an itch to go looking for a summary vid for the history of AI development milestones to remind me of those I’ve lived through & fill in any gaps & highlight the lesser known in the zeitgeist/ media highlighted players. Or a James Burke’s Pinball Effect / Connections type vid, that would be cool.
The first vid of a acceptable length at that moment that had some views claimed that computers were using tape drives in the ‘90s & so was either written by an AI or a 12yr old. So I quickly moved on and had other things to do.
Lo & behold a week later the Verge itself posts exactly what I was looking for from a known credible source that I could rely on being well researched & refer to academic papers !!!
Thx 🙏
What I learned from hanging with boys is that when their GF is talking just repeat the last word in their sentence and nod, and they think that they're already making a conversation. So basically this is how basic language models were developed.
No Jabberwackey or Cleverbot?
I remember playing with ELIZA in the early 80s. I even custom modified it to expand its responses.
yeeeeeeeeeeessss finally someone honors dr sbaitso!
Damn. You guys should see what I've put together over the last few days.. Got duelplex dialogue working as of tonight So you and bot can speak at the same time. TTS powered by Eleven Labs and technically no latency.
I first learned about ELIZA after playing with the chatbot on the promotional website for the movie AI.
It's beautiful to see her getting the light she deserves in this saga. She was always beautiful. Where are the haters now?
honestly, AI is quite good now.
Very interesting insight.
All I need is the Emacs doctor, and I'm planning to bring up in my next session how incredibly disappointed I was that this video made no mention of it. M-x doctor for life.
The description of transformers as “being able to see the entire sentence” is true about the type of model used in “Attention is all you need”, however the type of model used for ChatGPT and other similar chatbots *is* only predicting the next word (or token).
decoder only models
SmarterChild! Totally forgot about it!
"We're not at that magical moment..." Yeah that's 6 months from now
A little less actually.
Grew up with sbaitso on my thunder board way back in the day
10:30 for me, it's more bc i realize that chatgpt's responses aren't hardcoded, so it's not always going to give the correct answer
I instantly got reminded of the movie 'Her'.
An ad for this video came up during this video... 😂
More video content like this please... the Verge has had ever fewer pieces like this since Dieter left. In short, more Dieter please. Thank you.
I asked Gemini to impersonate ELIZA but its response to "all men are the same" was basically the same as ChatGPT's
Gus Fring as a bot would be scary.
Sorry to say, in a few years, ChatGPT might be seen as slow compared to newer, more effective tools.
How's the Verge and its editors not savvy enough to omit the phrase "ok Google"?
@0:57 How can there be a Mac in 1966? Apple wasn't even founded then.
It's totally not because chatGPT reads like a human that I forgive its mistakes. It's because I know that despite it sounding so authoritative and confident, it doesn't always get the data right.
8:05 Sorry but the explanation about transformers is misleading, not simplified. Depende models were a think way before transformers, so saying AI before couldn't see more than 1 word in the text is not simplifying, but just wrong.
Dark chatgpt
Chatgpt got 9 divided by 3 wrong. It seems awful at basic math. After I called it out it apologised and said I was right and that while it strives to get it right it doesn't always. Seriously.
Cleverbot came out 10 years ago and still has more personality than chatGPT
All Bots are alike
Where's simsimi?
Cleverbot!
❤❤❤
These really are robots in disguise
I've tried a couple of these bots. And I can't really find any reason to talk to them. They don't really feel like anything that interests me. I usually just end up insulting them in an attempt to get a rise out of them. And move on. Nothing to see here...
You lack imagination then because these new chatbots are engaging and actually useful.
The problem is the DEI parameters in Chatbot.
AT least when you divorce a bot. She it take half your money.
I played this at 2x speed and it was still so slow LOL. SPEAK FASTER DAVID.
it's not person
I don't see the appeal. It's a less efficient way of getting things done.
It's very good as a writing help.
@@paulu_ yeah that’s the only use I can think of. But I hate when companies use bots for customer service. They’re just there to annoy you until you find a human.
@@djp1234 what if the bot was so good you couldn't tell it wasn't a human?
@@87TechReviews never seen a bot that good. Every single bot I’ve interacted with was useless and couldn’t solve my problem.
@djp1234 it helps you find information and parse through the parts you didn't ask about. Think of it as an enhanced search engine. It's a huge productivity booster for administrative office positions, programming positions... just those alone will revolutionize society. Programmers already have done that without the autocoding chatbots so just imagine
very weak video
❤❤❤