Encyclopedia Britannica Is Now an AI Company 59
An anonymous reader quotes a report from Gizmodo: Once an icon of the 20th century seen as obsolete in the 21st, Encyclopedia Britannica -- now known as just Britannica -- is all in on artificial intelligence, and may soon go public at a valuation of nearly $1 billion, according to the New York Times.
Until 2012 when printing ended, the company's books served as the oldest continuously published, English-language encyclopedias in the world, essentially collecting all the world's knowledge in one place before Google or Wikipedia were a thing. That has helped Britannica pivot into the AI age, where models benefit from access to high-quality, vetted information. More general-purpose models like ChatGPT suffer from hallucinations because they have hoovered up the entire internet, including all the junk and misinformation.
While it still offers an online edition of its encyclopedia, as well as the Merriam-Webster dictionary, Britannica's biggest business today is selling online education software to schools and libraries, the software it hopes to supercharge with AI. That could mean using AI to customize learning plans for individual students. The idea is that students will enjoy learning more when software can help them understand the gaps in their understanding of a topic and stay on it longer. Another education tech company, Brainly, recently announced that answers from its chatbot will link to the exact learning materials (i.e. textbooks) they reference.
Britannica's CEO Jorge Cauz also told the Times about the company's Britannica AI chatbot, which allows users to ask questions about its vast database of encyclopedic knowledge that it collected over two centuries from vetted academics and editors. The company similarly offers chatbot software for customer service use cases. Britannica told the Times it is expecting revenue to double from two years ago, to $100 million.
Until 2012 when printing ended, the company's books served as the oldest continuously published, English-language encyclopedias in the world, essentially collecting all the world's knowledge in one place before Google or Wikipedia were a thing. That has helped Britannica pivot into the AI age, where models benefit from access to high-quality, vetted information. More general-purpose models like ChatGPT suffer from hallucinations because they have hoovered up the entire internet, including all the junk and misinformation.
While it still offers an online edition of its encyclopedia, as well as the Merriam-Webster dictionary, Britannica's biggest business today is selling online education software to schools and libraries, the software it hopes to supercharge with AI. That could mean using AI to customize learning plans for individual students. The idea is that students will enjoy learning more when software can help them understand the gaps in their understanding of a topic and stay on it longer. Another education tech company, Brainly, recently announced that answers from its chatbot will link to the exact learning materials (i.e. textbooks) they reference.
Britannica's CEO Jorge Cauz also told the Times about the company's Britannica AI chatbot, which allows users to ask questions about its vast database of encyclopedic knowledge that it collected over two centuries from vetted academics and editors. The company similarly offers chatbot software for customer service use cases. Britannica told the Times it is expecting revenue to double from two years ago, to $100 million.
Re: (Score:3)
"Valued at..." Which means that the moment that someone tries to collect some of that and make it money, the value will drop.
Re: (Score:3)
Because, of course, the only facts worth knowing are the ones that were discovered in the last financial quarter, or something.
Re:Wikipedia (Score:4, Informative)
Possibly. Possibly not. A quality encyclopedia is better at editing than Wikipedia, and leaves out a lot of utterly useless stuff which otherwise forms the bulk of Wikipedia. Also the references in Encyclopedia Britannica actually exist, unlike the collection of broken web links in all the Wikipedia articles.
Skeptical (Score:4, Informative)
The tendency to rewrite history and basic scientific articles to fit a political narrative since 1980 is a good reason to be skeptical.
I'd like a critical and reasoned discussion of contemporary topics and not just using a large set of agenda based research to prevent discussion of encyclopedia articles.
Research sources for encyclopedia articles need to be evaluated for scientific quality (no self-reported surveys, AI sentiment analysis, or microfraction of a group's social media posts), freedom from an overt agenda based bias and, in turn, not be based on earlier sources of questionable scientific quality.
Encyclopedia article authors also need to have proven as a knowledgeable person in the subject and not a producer of agenda based research or agenda based books.
And the building blocks of encyclopedia articles should not be solely based on 20 year research studies of the social status, views, political beliefs or trends of a prior generation. More modern research should be done to reevaluate the outdated scientific research especially in political topics or social sciences.
Re: (Score:2, Insightful)
The tendency to rewrite history and basic scientific articles to fit a political narrative since 1980 is a good reason to be skeptical.
Tell me, what's so good about the blatant biases and political narrative of the 50s, 60s and 70s that you wish for them to be set in stone forever?
Re: (Score:2)
History doesn't change, but our information about it does. And saying history shouldn't "change" after the consensus changes in the 1980s that maybe it's bad that facts were written out of history because they were politically inconvenient for literally a hundred years is being in favor of the rewriting of history in the 1870-1950s, not being in favor of it being set in stone.
Re: (Score:2)
My original point: I'd like a critical and reasoned discussion of contemporary topics and not just using a large set of agenda based research to prevent discussion of encyclopedia articles.
Examine a history article, the original sources it is based on, the historian authors, the other research by the historian authors and let those be examined for agenda based research so that agenda based research is reevaluated and possibly removed as a basic or 'fact' in the history article.
For example, examining a his
Re: (Score:2)
Adding
- And excluding the most modern "use an AI with a data set" to conclude proposition A is true when that same research cannot be reproduced in the future because, the data set is frozen in time, impossible to train another AI the exact same way, and that there are no general scientific equations derived from the AI's association of input to output.
Britannica does that old timey fact checkin thing (Score:3)
Wikipedia has made every enclocpieda obsolite
No. Britannica is fact checked far beyond Wikipedia. It would probably be somewhat safe to let an AI learn from it, unlike Wikipedia.
Re: (Score:2)
Britannica made mistakes and is out of date for a lot of things. Not a good idea to let AI learn from it.
Britanica is not stuck at its last printed 2010 version, it's gone digital. Try their chat bot, it seems pretty current. Their "New on Britannica" shows they are still adding topics.
When playing with the chatbot note whether it is showing an answer from Britannica or Chat-GPT. If their system doesn't come up with an answer they seem to check Chat-GPT.
Re: (Score:1)
I'm sorry, but it seems like your request is beyond the scope of what I can assist with. If you have any other questions or need information on a different topic, feel free to ask!
thats what i got....
Re: (Score:2)
"There are no American troops in Baghdad!". "Hey Encyclopedia Brittanica, where there American troops in Baghdad on April 8th 2003". "No. According reliable sources, the ministry of information of Iraq, there were no American troops in Baghdad on April 8th 2003".
Britanica answered your question correctly (Score:2)
It is totally useless for AI to "learn from it" because it will not gain any new knowledge, which an encyclopedia should. Google has the algorithm to scrape EB and tell me what's in it. No AI can enter new knowledge. "There are no American troops in Baghdad!". "Hey Encyclopedia Brittanica, where there American troops in Baghdad on April 8th 2003". "No. According reliable sources, the ministry of information of Iraq, there were no American troops in Baghdad on April 8th 2003".
No, Britanica actually answered correctly.
"where there American troops in Baghdad on April 8th 2003
Sources:
Iraq War: The 2003 conflict
Iraq: Post-9/11 tension and the U.S.-led invasion of Iraq
Baghdad: Baghdad in modern Iraq
This Day in History: April 9
Yes, American troops were in Baghdad on April 8, 2003. U.S. forces had advanced into the city and were conducting operations in the heart of Baghdad around that time. By April 9, resistance in Baghdad had collapsed, and U.S. soldiers took control of t
That's a bit of a shift (Score:2)
Moving from a business with products based on facts, to one that produces hallucinations?
Re: (Score:2)
That has helped Britannica pivot into the AI age, where models benefit from access to high-quality, vetted information. More general-purpose models like ChatGPT suffer from hallucinations because they have hoovered up the entire internet, including all the junk and misinformation.
They have a brave interpretation of the hallucination problem, that's for sure, and it's doing a lot of work here. My call is that AI hallucinates because "colorless green ideas sleep furiously", but well, one has to try. It's not like things can get much worse for them.
Re: (Score:2)
While I agree that without true understanding LLLMs are going to hallucinate some of the time, the sheer amount of nonsense used to train ChatGPT and Gemini must be a problem if accuracy is important in a response to a prompt.
Re: (Score:3)
"Hallucination" is the wrong word. It is based upon the idea that there is "intelligence", of which none exists. There's very advanced and capable pattern matching, but not intelligence. Better to just call it a bug for what it is. Or a probabalistic outlier.
Re: (Score:2)
If you want to be nitpicky, "delusion" would be a better word. But yes, that implies *belief*, however philosophically that implies a lot of stuff happening under the covers that isn't there -- a model of reality. As I understand it what the model does is generate a plausible looking response, which is something people do too, but it's not like have an actual belief.
Re: (Score:2)
Maybe it's about improving the quality of the hallucinations. Android dreams and electric sheep and all that.
Re: (Score:2)
Moving from a business with products based on facts, to one that produces hallucinations
Not quite. The Britannica's LLM is trained on the sum total of the encyclopedia's articles and media, ChatGPT is trained on reddit. I think Britannica is correct in thinking that "hallucinations" will not be a problem with their product - as long as they can train it to recognize what lies outside if its knowledge base.
I'm glad Brittanica survived (Score:4, Insightful)
Re:I'm glad Brittanica survived (Score:4, Interesting)
If they cut out the crap they have enough money to indefinitely fund their hosting off interest earned. Per their own website, they spend over twice as much paying for the fundraising efforts as they do hosting. I suspect they could survive just fine at only 5% of their budget. The other 95% is just because they have the money and they can.
Re: (Score:2)
Salaries aren't necessary? Even if they get content for free, there's still a lot of work maintaining a site that serves up eighteen billion page views a month.
Re: (Score:2)
Also I suspect that the majority of references in Wikipedia are broken.
We had World Book growing up, which was sold door to door when I was a toddler. That was a piece of crap. Even when I was ten I realized that it was woefully out of date and not very well written and that it had a whole lot of opinions.
Re: (Score:2)
I've got a collection of paper encyclopedias that I aquired just as Wikipedia was overtaking the rest of the encyclopedia market. AI and websites may be "smart", but a book can last for centuries.
You're right. A book can last for centuries. But printing all of them on paper, cannot. We were half the current population less than half of one century ago. Not sure why you’re talking in centuries. I doubt consuming trees for that purpose will remain legal past this century. For billions more reason.
A quick test... it's not very helpful. (Score:2)
ChatGPT: What are non-market practices?
Answer: a long description of all the various aspects about non-market practices.
Britannica: What are non-market practices?
Sorry, the Britannica knowledge base does not contain this information.
Re: (Score:2)
Given the last printed edition was revised in 1974, you wouldn't find "Internet" in it either.
Wikipedia says 2010 last printed version (Score:2)
Given the last printed edition was revised in 1974, you wouldn't find "Internet" in it either.
Wikipedia says: "Encyclopædia Britannica ... The 2010 version of the 15th edition, which spans 32 volumes and 32,640 pages, was the last printed edition. Since 2016, it has been published exclusively as an online encyclopaedia."
Re: (Score:3)
So it knows nothing since 2010. Let's train an AI on more than a decade old information...
Wrong. They continue to work on the encyclopedia. They just don't print it anymore, it's an online service. They offer a sampling of things recently added.
https://www.britannica.com/new... [britannica.com]
Re: (Score:2)
they finished updating the printed version in 2010 like you said
No, I did not say they finished updating, I said it was the last printed version. I also said it continued as an online version, and provided a linked to their "New Articles" page.
Re: (Score:2)
yes training AI on the 2010 version would be bad. Glad you now agree.
Wrong. I never mentioned a date with respect to training. A reply stated the wrong date for the print version and I corrected that. No one mentioned limiting training to 2010 except you.
Re: (Score:2)
great that you now agree the training of AI with decade+ outdated information is not a good idea.
Wrong. There was no change. I always supported training with current data, which Encyclopedia Britannica has always provided.
Re: (Score:2)
No change? Then you agreed all along that decades old data is bad?
I never said old data was OK, that was just an AC's bad guess.
Re: (Score:2)
they don't continue to work on the printed version
Which is why I referred to their continued work on the online version.
Re: (Score:2)
See decade old data is bad for an AI
Good thing I never suggested using old data, that was an AC's bad guess.
A quick test ... you are clueless (Score:2)
Britannica: What are non-market practices?
Sorry, the Britannica knowledge base does not contain this information.
Actually, Britannica answers with:
"Non-market practices refer to activities and systems that operate outside the traditional market framework, where goods and services are not exchanged through buying and selling using money. These practices can include barter systems, ecosystem services, and government interventions to address market failures.
Barter Systems: Barter involves the direct exchange of goods or services without using money as an intermediary. It is one of the oldest forms of commerce and i
Re: (Score:2)
That is just bullshit (Score:2)
Unless they plan on making really crappy content now, regular search is all you need for their stuff.
Excellent! (Score:3)
Re: (Score:3)
Coming back for this (Score:2)
A boon for the generations toostupid to read (Score:2)
Britannica has been around for over two hundred years. I sold a LOT of sets i my day. Unfortunately, there are huge numbers of people who can't or don't read. In my old age, I find myself tutoring/coaching people on how to think and how to study. This last year, I have been coaching 4 people, WITH MULTIPLE DEGREES, who read only about 150 words-per-minute and can't remember what they read. (In my generation , boomer, the average reading speed was 200-250 wpm w/70% comprehension.) A report by abtaba (https:/