I keep seeing you everywhere and saying hiii but now I realize it looks like I'm Internet stalking you😅 we just watch the same videos lol Also hii lexiiii
Hi Lexi, great seeing you here! It would be good if it didn't take 10 minutes to get it to give wrong information. literally just asked it what black Sand is made of and it didn't even mention basaltic Rock! The main component!!! Not to mention that the underlying LLM that's doing the Summary is probably just as unethically trained as any other.
@@tinpin609 That's not how AI works. AI uses set training data, meaning it's not live-reading anything, rather you feed it a 'version' of something and it learns off of that. For example, ChatGPT 3.0 can't reference anything after 31 October 2023, as that's the endpoint.
@@SeanORaighconsidering the energy usage of your house alone, this is far in the acceptable territory. You just desperately need an excuse to frame AI negatively
ok but this makes sense. The ai is not changing the encyclopedia, it’s just training on possibly the largest historical archive on the planet. It just makes sense, and will only make AI smarter and less stupid (ex. Saying that 1920 was 30 years ago)
All SOTA models already have all that knowledge, Britannica's is just fine tuned specifically for their own content without the entirety of the scraped internet on top. It's not going to make AI smarter, it simply makes their AI specialized. It can still hallucinate and fuck up like any other model if you prompt it just the right way, which people are bound to do by accident or otherwise. Trash-in-trash-out only goes so far. You can have prestine content to train on and still have a model that's acts like a dipshit sometimes.
this is what we call an Expert System. You're just asking it to retrieve something from a large database as a response to a voice or text prompt from you. it doesn't have to make anything up, it's just a very effective way to point you to what you want, which is how this industry SHOULD HAVE STARTED.
This whole industry is a scam anyway, most of the models being used are just chatbots with extra steps, and much like that Amazon AI store that got proven to only be an office in India, I'd wager a solid chunk of other "AI" models are Actually Indians, or other people working in some other horrifically underpaid outsourcing office.
It's kind of hilarious that AI keeps being shilled for all is most garbage, dysfunctional uses, while it is actually pretty good as a "mega search engine". Unfortunately the more ridiculous uses get more investor cash (because investors are stupid as bricks), and the search engine use would need to seriously address the copyright concerns. So it's all garbage for now.
I'm not angry about people using training data they actually do own and aren't relying on mass entrapment to generate. I am concerned about the way LLM machines work and how likely it will be to manufacture facts.
It is literally the best use I could think of. Just an all knowing, reliable and ethical source of knowledge. I can't think of a reference for it, but there's probably something out there
@@danielsjohnson But that should be heavily mitigated, since the AI is being trained on the sum of all human knowledge as presented by Britannica. Not being taught everything on the internet.
This sounds like a good idea. And honestly, I'd sooner trust EB than Wikipedia if it was the info within was as easy to access as Wiki. An AI to hunt down and summarize the relevant pages for you sounds like a positive step. (To be clear, I wouldn't stake my life on either without checking the sources lol)
This is one of the few ai stuff that i might be fine with. Since its taking stuff from these book. Not just random crap on the Internet. Still not a fan of ai but this is one of the few that might be fine.
How are they going to power it? Maybe think about it for more than a single second before bending over to glaze this vile company for jumping onto a grift that is literally speeding us toward extinction.
AI is great, what do you mean? It's ability to create smut both in art and in writing has changed my life, don't even get me started on the AI Roleplayers and Chatbots
@@coastaku1954ai is great when implemented and used with care. It will have a huge effect on our education system that we don’t yet have the answers for (education all over the world needed a shake-up anyway, but sad it had to come to this). Children in school today will probably get worse learning until the system catches up, which it hopefully will. Makes me sad for the current generation of school children
with all due respect having an AI that has the knowledge of a load of encyclopedias actually sounds practical and optimistic for the future quality of life for humans, compared to the AI generated art slop or even normal literature, encyclopedias are meant to provide knowledge as effective and comprehendible as possible, and an AI assistant in this case isn't just a garbage marketing label but might be a flexible tool to ask "how do you make a clove knot with a 2m rope"
I love that when showing the impressive history of the worlds largest physical encyclopedia, you show a screenshot from the world's largest encyclopedia, Wikipedia.
to be fair, its database would be made of what possibly the most exhautive and reliable source in the world. I hope they don't fck this up and leave any chance for the AI to hallucinate...
This is genius. Britannica AI won't be able to go rogue and start learning from AI like other models do, which is why it won't spew out nonsensical slop when asked simple questions.
This is an acceptablenuse of AI. It trained on its own data, and isn't acraping data from thr internet. So no risk of degradation, corruption and deciation. The AI copycats that will spawn from this, though? Yeah nah.
As long as the AI Ais trained only on the Britannica information then this might be good. That being said, I hope they lose money on this out of spite.
Out of spite for what? That makes no sense. Hoping they fail is like saying you want a chef to burn down their kitchen because they’re trying a recipe you don’t like. Specialized, fine-tuned, ethical AI is one of the few things keeping this tech from turning into a complete shitshow. Instead of rooting for failure, you should be rooting for them to succeed and make enough money to show the market there’s value in doing this the right way. Because make no mistake about it, AI is forever now, and if people wanna see it go the right way, they will support those who do it right.
^^^ this is smart use of AI. Trained on highly controlled, highly factual data. Already owned by the same company. Vast amount of text. Perfection. Not slop.
Honestly a smart way to get people to the use the knowledge in those books. Let’s be real, those things are so dense and there’s so many most people wouldn’t even consider buying one
It’s actually very nice to hear about an AI company that didn’t steal a bunch of stuff to train their programs and instead used their own work to do so
I remember several sets of encyclopedias in the house as a kid. As well as the program Microsoft ENCARTA. A digital encyclopedia where I did some elementary school research on before we got internet.
The comments saved this one, because I was about to hop on here with fury in my fingers ready to type up a storm. This feels like we finally discovered the only ethical way to run an AI business and is probably the only ethical AI company in existence.
If their AI is just digital librarian pointing you to what you want to find out then this would also make them the first actually useful AI beyond just novelty...
THIS is the kind of AI I'm fine with. They used their own material, to train an AI model. They didn't rob anyone, well if they did, it's lost to time. And frankly that may be the best model so far. Unless they start robbing other stuff from people like general AI does
You know i really like this, a real source of knowledge being feed to an ai is a very good idea to me, now it would be a bad move if the ai can learn from the public
If its anyone that should be making Chatbots, its the people while near unlimited access to the vast majority of human knowledge they can train the bots on without stealing data
A LLM trained EXCLUSIVELY on a well respected encyclopedia, by the owners of that encyclopedia? Yeah, I'm actually down for that. If they fix the "hallucination" problem, this is a perfect reference tool.
Cool that they’re training it on their extensive dataset but chatgpt/claude are 99% gonna still be a better experience for the end user at the end of the day
12 ชั่วโมงที่ผ่านมา
I mean...thats trully a great development in the use of AI: as a tool for assistance
Okay, lemme say this: I absolutely despise, loathe, hate and scorn artificial intelligence with an unyielding, unfathomable passion. I'm an artist, go figure. And I've not ONCE used Chat GPT ever, except for that one brief moment I tried to get it to hate itself out of curiosity. But this? Honestly, I fuck with that. THIS is what AI should be used for, goddammit! Human knowledge AND actually *_PROPERLY CURATED_* human knowledge that *_is NOT_* simply skimmed off the internet randomly, stealing and plagiarising content in the process and leading to outrageous amounts of misinformation. This is one of the good ones, I dare say... at least I hope. Noice. Only problem is that it's most likely still built on one of those open source AI models that were already trained on stolen information, anyway, so... you really can't have nice things under capitalism, that's just fundamentally incompatible at this point, but I guess we can celebrate the tiny sparks of un-shittiness that still manifest _in spite of it,_ every now and then.
@@Pharozos Ah yeah thanks... just saw his 2019 video explaining why he moved to SA. Never knew! Glad he is making a positive influence to the community while he is here as well!
This is the one and only time i'm gonna say this doesn't sound so bad, if they only fed it info from the encyclopedias it could work, that's way more limited than the entirety of the internet.
I mean, given it's basically gonna be a fancy search engine for Encyclopedia Britannica's vast collection of articles and documentation, I don't mind it. Plus, it will be more trustworthy than ChatGPT or Gemini.
the main problem with these AI's is that they're predictive language models, they don't actually know anything, they just spit out what words are most likely to come next as such, they will make things up, even after being trained on true information. It still needs a lot of refinement before I'd consider it trustworthy
an ai that lets you sift through and find out what you need from the metric tons of encyclopedia’s? One that is trained off ethically sourced material? Sounds like a w idk
I had the expanded Encyclopedia Britannica and I actually read the whole thing. So, if just having it makes you a Giga-nerd then what does that make me?
it's not very much of a problem until you realize that there's a lot of incorrect information on britannica itself (for example, it lists Praesepe as a constellation)
Non-stolen work and a model that should be able to provide references? So an actually useful AI that has all rights to its training data? Fantastic.
Hey, didn't expect to see you here, Lexi!
Lexi!
hai :3
I keep seeing you everywhere and saying hiii but now I realize it looks like I'm Internet stalking you😅 we just watch the same videos lol
Also hii lexiiii
Hi Lexi, great seeing you here!
It would be good if it didn't take 10 minutes to get it to give wrong information.
literally just asked it what black Sand is made of and it didn't even mention basaltic Rock! The main component!!!
Not to mention that the underlying LLM that's doing the Summary is probably just as unethically trained as any other.
Ngl kinda hope this works considering it's ethically AND reliably sourced, as opposed to almost all other AIs
Yeah one of the few times I’m happy to hear about AI
Agreed
Hope so but I doubt it. I doubt there's enough data there for it to be a good AI model
And it sounds epic on top
The energy consumption alone makes this ridiculously wrong.
AI based on a legit, validated, accurate encyclopedia instead of the terror that is the internet is a win IMO.
It can cite ChatGPT instead of Britannica
Yeah, there’s no way anyone could just go in and edit it secretly. Digital platforms are impervious after all. /s
I agree with ya!
@@tinpin609 That's not how AI works.
AI uses set training data, meaning it's not live-reading anything, rather you feed it a 'version' of something and it learns off of that. For example, ChatGPT 3.0 can't reference anything after 31 October 2023, as that's the endpoint.
AI has some of the best answers to the questions I ask on google, it’s a god send for information.
Thinking about it, this is perfectly acceptable ai usage. They arent taking other people's work or anything like that.
True
Considering the energy usage, this is about as far from acceptable as it's possible to be.
@@SeanORaigh Hench why they are pushing for more Nuclear Power now.
Id still say lesser evil @@SeanORaigh
@@SeanORaighconsidering the energy usage of your house alone, this is far in the acceptable territory. You just desperately need an excuse to frame AI negatively
ok but this makes sense. The ai is not changing the encyclopedia, it’s just training on possibly the largest historical archive on the planet. It just makes sense, and will only make AI smarter and less stupid (ex. Saying that 1920 was 30 years ago)
Yeah, most commenters failed to understand this
All SOTA models already have all that knowledge, Britannica's is just fine tuned specifically for their own content without the entirety of the scraped internet on top. It's not going to make AI smarter, it simply makes their AI specialized. It can still hallucinate and fuck up like any other model if you prompt it just the right way, which people are bound to do by accident or otherwise. Trash-in-trash-out only goes so far. You can have prestine content to train on and still have a model that's acts like a dipshit sometimes.
Yeah, i think this is one of the best uses of ai I've seen. No false data. Nothing that doesn't have a source. Just solid info, streamlined.
allat just for it to still not know how many 'R's are in strawberry
Bruh AI that is trained on factually correct data still will give false answers. It is not dependable.
this is what we call an Expert System. You're just asking it to retrieve something from a large database as a response to a voice or text prompt from you. it doesn't have to make anything up, it's just a very effective way to point you to what you want, which is how this industry SHOULD HAVE STARTED.
Kinda sad openai didn’t think about this nearly enough, if at all
This whole industry is a scam anyway, most of the models being used are just chatbots with extra steps, and much like that Amazon AI store that got proven to only be an office in India, I'd wager a solid chunk of other "AI" models are Actually Indians, or other people working in some other horrifically underpaid outsourcing office.
It's kind of hilarious that AI keeps being shilled for all is most garbage, dysfunctional uses, while it is actually pretty good as a "mega search engine".
Unfortunately the more ridiculous uses get more investor cash (because investors are stupid as bricks), and the search engine use would need to seriously address the copyright concerns. So it's all garbage for now.
Finally, an Ai application that actually makes sense and one that will actually be reliably helpful.
Yeah, I assume this is just an easier way for users to navigate the 30+ book series instead of having to endlessly comb through the pages for hours
Wrong on both counts.
@ please explain because you clearly don’t understand how Ai works
Oh you do? XD there are no ais, those are llms and are shit :D
@@psygonzo7974 as if anyone cares about the difference at this point
I mean training AI on own data, couldn’t be the worst
I'm not angry about people using training data they actually do own and aren't relying on mass entrapment to generate. I am concerned about the way LLM machines work and how likely it will be to manufacture facts.
It is literally the best use I could think of.
Just an all knowing, reliable and ethical source of knowledge.
I can't think of a reference for it, but there's probably something out there
god
Its good, but you still need to worry about the "confidently incorrect" mistakes that all AIs make.
@@danielsjohnson But that should be heavily mitigated, since the AI is being trained on the sum of all human knowledge as presented by Britannica. Not being taught everything on the internet.
@@HollowBagel Should it be? Ideally yes. Will it be? I don't know. Time will tell.
@@danielsjohnsonfair, but fingers crossed
This sounds like a good idea. And honestly, I'd sooner trust EB than Wikipedia if it was the info within was as easy to access as Wiki. An AI to hunt down and summarize the relevant pages for you sounds like a positive step. (To be clear, I wouldn't stake my life on either without checking the sources lol)
This is one of the few ai stuff that i might be fine with. Since its taking stuff from these book. Not just random crap on the Internet. Still not a fan of ai but this is one of the few that might be fine.
How are they going to power it? Maybe think about it for more than a single second before bending over to glaze this vile company for jumping onto a grift that is literally speeding us toward extinction.
AI is great, what do you mean? It's ability to create smut both in art and in writing has changed my life, don't even get me started on the AI Roleplayers and Chatbots
@@coastaku1954ai is great when implemented and used with care. It will have a huge effect on our education system that we don’t yet have the answers for (education all over the world needed a shake-up anyway, but sad it had to come to this). Children in school today will probably get worse learning until the system catches up, which it hopefully will. Makes me sad for the current generation of school children
with all due respect having an AI that has the knowledge of a load of encyclopedias actually sounds practical and optimistic for the future quality of life for humans, compared to the AI generated art slop or even normal literature, encyclopedias are meant to provide knowledge as effective and comprehendible as possible, and an AI assistant in this case isn't just a garbage marketing label but might be a flexible tool to ask "how do you make a clove knot with a 2m rope"
This is FINE. It's like when the future museums have a hologram guide.
Honestly, this is what AI should be (if it even works as intended); offering an amalgamation of knowledge from a trusted dataset.
I love that when showing the impressive history of the worlds largest physical encyclopedia, you show a screenshot from the world's largest encyclopedia, Wikipedia.
AI trained on very specific stuff is so useful, i hope this works out for them.
to be fair, its database would be made of what possibly the most exhautive and reliable source in the world. I hope they don't fck this up and leave any chance for the AI to hallucinate...
This is genius. Britannica AI won't be able to go rogue and start learning from AI like other models do, which is why it won't spew out nonsensical slop when asked simple questions.
you mean next year, right? this year has like three days left
lol
This is an acceptablenuse of AI. It trained on its own data, and isn't acraping data from thr internet. So no risk of degradation, corruption and deciation.
The AI copycats that will spawn from this, though? Yeah nah.
As long as the AI Ais trained only on the Britannica information then this might be good. That being said, I hope they lose money on this out of spite.
You hope pretty much the only ethically trained AI loses its creator money?
@7HEMUFFINMAN yeah, why not
@@riotbreaker3506 understandable, have a nice day
@@riotbreaker3506 does that make you ethically correct, no? we need more ethical ai.
Out of spite for what? That makes no sense. Hoping they fail is like saying you want a chef to burn down their kitchen because they’re trying a recipe you don’t like. Specialized, fine-tuned, ethical AI is one of the few things keeping this tech from turning into a complete shitshow. Instead of rooting for failure, you should be rooting for them to succeed and make enough money to show the market there’s value in doing this the right way. Because make no mistake about it, AI is forever now, and if people wanna see it go the right way, they will support those who do it right.
^^^ this is smart use of AI. Trained on highly controlled, highly factual data. Already owned by the same company. Vast amount of text. Perfection. Not slop.
Obviously it'll still invent stuff out of nowhere, but it'll probably do it way less
AI generated Encyclopaedia Britannica 💀
Honestly a smart way to get people to the use the knowledge in those books. Let’s be real, those things are so dense and there’s so many most people wouldn’t even consider buying one
Bruh we cant lose it now !
You could accomplish the same thing with a fully digitized copy of the books and a search function, except without the hallucinations
Finally someone said it!
It’s actually very nice to hear about an AI company that didn’t steal a bunch of stuff to train their programs and instead used their own work to do so
Well, I'll stay a while longer with the Hitchhiker's Guide, and wait until I'll get copy of the Encyclopedia Galactica. Totally woth it.
That is genuinely the best to use for AI I've seen so far.
I have a 1997 unabridged edition. I keep it to prove people are changing definitions lol.
this is actually a really good and ethically correct use of ai
Why does bro have the green m&m on his shelf
this is honestly a pretty sick way to use ai, and now i kinda wanna make an open source private ai based on Britannica ai and everything in wikipedia
I just asked it how many bones a cat has and the source was ChatGPT. It wasn't even their own encyclopedias
I remember several sets of encyclopedias in the house as a kid. As well as the program Microsoft ENCARTA. A digital encyclopedia where I did some elementary school research on before we got internet.
Seems like an actually decent use of ai
This is 100% a perfect use of AI 🎉
Very good for them for adapting to the changing times
When I was 6 I had an almost complete set. I still don’t know at 34 who bought it for me but BABY!!! I cherished those thangs
The comments saved this one, because I was about to hop on here with fury in my fingers ready to type up a storm. This feels like we finally discovered the only ethical way to run an AI business and is probably the only ethical AI company in existence.
Finally an ai trained properly and ethically! Trained on nothing but facts and proper knowledge!
This sounds pretty solid tbh. It's just being trained using the data set of encyclopedia brittanica? Fabulous, in this instance.
The best thing about AI is that the craxy power Consumption is pushing the deregulation of Nuclear Power.
Damn being able to talk with books from my childhood is never something I considered possible
Wow! High quality model based on high quality data that wasn't stolen.
If their AI is just digital librarian pointing you to what you want to find out then this would also make them the first actually useful AI beyond just novelty...
THIS is the kind of AI I'm fine with.
They used their own material, to train an AI model.
They didn't rob anyone, well if they did, it's lost to time.
And frankly that may be the best model so far. Unless they start robbing other stuff from people like general AI does
Nice shirt you got there.🇿🇦
A giant hand crafted training data set like that is worth a lot.
"Britannica AI" sounds like the villain from a sci-fi property that hasn't been made yet
You know i really like this, a real source of knowledge being feed to an ai is a very good idea to me, now it would be a bad move if the ai can learn from the public
If its anyone that should be making Chatbots, its the people while near unlimited access to the vast majority of human knowledge they can train the bots on without stealing data
damm i didnt know britanica was that old
A LLM trained EXCLUSIVELY on a well respected encyclopedia, by the owners of that encyclopedia? Yeah, I'm actually down for that. If they fix the "hallucination" problem, this is a perfect reference tool.
Arguably this is the best use for AI, cataloging and making information more accessible :)
This is a huge game changer.
I hope New York Times is next, what a fabulous archive to train on.
Cool that they’re training it on their extensive dataset but chatgpt/claude are 99% gonna still be a better experience for the end user at the end of the day
I mean...thats trully a great development in the use of AI: as a tool for assistance
Okay, lemme say this:
I absolutely despise, loathe, hate and scorn artificial intelligence with an unyielding, unfathomable passion. I'm an artist, go figure.
And I've not ONCE used Chat GPT ever, except for that one brief moment I tried to get it to hate itself out of curiosity.
But this?
Honestly, I fuck with that.
THIS is what AI should be used for, goddammit! Human knowledge AND actually *_PROPERLY CURATED_* human knowledge that *_is NOT_* simply skimmed off the internet randomly, stealing and plagiarising content in the process and leading to outrageous amounts of misinformation.
This is one of the good ones, I dare say... at least I hope.
Noice.
Only problem is that it's most likely still built on one of those open source AI models that were already trained on stolen information, anyway, so... you really can't have nice things under capitalism, that's just fundamentally incompatible at this point, but I guess we can celebrate the tiny sparks of un-shittiness that still manifest _in spite of it,_ every now and then.
Parents had a 70s brittanica and I read all 34 or so volumes a few times😅
Finally an AI company that didn't steal training data.
Bro! Are you a Springboks fan? This is the second time that I see you in Boks gear.
He is from SA so it checks out.
@@PharozosI hope noone thought of the other SA
@@Pharozos Ah yeah thanks... just saw his 2019 video explaining why he moved to SA. Never knew! Glad he is making a positive influence to the community while he is here as well!
@@itsarian. South America? Saudi Arabia? South Australia? Schengen Area?
@clintonbadenhorst9082 no
This is the one and only time i'm gonna say this doesn't sound so bad, if they only fed it info from the encyclopedias it could work, that's way more limited than the entirety of the internet.
This can only lead to amazing things. Right?
Well Well Well
I mean, given it's basically gonna be a fancy search engine for Encyclopedia Britannica's vast collection of articles and documentation, I don't mind it. Plus, it will be more trustworthy than ChatGPT or Gemini.
This sounds about the first useful AI made by a company.
The data should be accurate.
Some companies are using AI art to make jigsaw puzzles. I’m not kidding.
The thing is, it still doesn't have knowledge. It's still just giving you the most likely words to follow the previous
THIS IS WHAT AI IS SUPPOSED TO BE USED FOR EVERYONE
the main problem with these AI's is that they're predictive language models, they don't actually know anything, they just spit out what words are most likely to come next
as such, they will make things up, even after being trained on true information. It still needs a lot of refinement before I'd consider it trustworthy
oh,god no.
Oh god yes. This is the perfect use for AI. Instead of ai art, we basically have a robot historian.
THIS YEAR?? it’s like not even a week anymore damn
I used to think the geek in the commercials was a hottie biscotti.
That's kinda sad, I learned English trying to read Britannica at the local public library before the internet was a thing.
an ai that lets you sift through and find out what you need from the metric tons of encyclopedia’s? One that is trained off ethically sourced material?
Sounds like a w idk
Honestly this is one of those rare times when I’m fine with ai
Still have my '99 full set of Britticana
finally, an actually purposeful ai with a good dataset
unlike those half baked quasi porno shits that force you to pay
I’m gonna have to think about this for a minute.
I had the expanded Encyclopedia Britannica and I actually read the whole thing. So, if just having it makes you a Giga-nerd then what does that make me?
its nice that this ai is trained on info from the Encyclopedia, but how are they avoiding hallucinations?
I'm fine with this
Data is theirs to do whatever they want with
I never understood how having an encyclopedia that disappears if the internet disappears made any sense.
I real they should collab with an ai so that it could tell you the definition of any word if you give it the prompt
we got reliably and ethically sourced ai before gta6
Oh, I thought it was something much, much worse. I hope this succeeds
that's deeply unfortunate
Encyclopedia Britannica will just start quoting Reddit posts.
Can you still get the latest physical edition?
This is how AI should be used.
Ok, this does make sense
#BoycottAI
But why?
@@notusneo because it replaces loser dead end jobs and then people who have no real purpose in life will start to question what they are doing here
@@sleepy.trader exactly.
I think it’s great, it will increase access and reduce price , so why not ?
Im glad it was sourced good, but it still feels super weird
they're using their own data, so it's good.
Why does an encyclopedia got Britain in it? Is it also in favor of slavery and colonialism?
Just keep whatever AI is training on it away from the rest of the internet and it can actually be a useful research tool.
it's not very much of a problem until you realize that there's a lot of incorrect information on britannica itself (for example, it lists Praesepe as a constellation)