Why YouTube Could Give Google an Edge in AI (theinformation.com) 30
Google last month upgraded its Bard chatbot with a new machine-learning model that can better understand conversational language and compete with OpenAI's ChatGPT. As Google develops a sequel to that model, it may hold a trump card: YouTube. From a report: The video site, which Google owns, is the single biggest and richest source of imagery, audio and text transcripts on the internet. And Google's researchers have been using YouTube to develop its next large-language model, Gemini, according to a person with knowledge of the situation. The value of YouTube hasn't been lost on OpenAI, either: The startup has secretly used data from the site to train some of its artificial intelligence models, said one person with direct knowledge of the effort. AI practitioners who compete with Google say the company may gain an edge from owning YouTube, which gives it more complete access to the video data than rivals that scrape the videos. That's especially important as AI developers face new obstacles to finding high-quality data on which to train and improve their models. Major website publishers from Reddit to Stack Exchange to DeviantArt are increasingly blocking developers from downloading data for that purpose. Before those walls came up, AI startups used data from such sites to develop AI models, according to the publishers and disclosures from the startups.
The advantage that Google gains in AI from owning YouTube may reinforce concerns among antitrust regulators about Google's power. On Wednesday, the European Commission kicked off a complaint about Google's power in the ad tech world, contending that Google favors its "own online display advertising technology services to the detriment of competing providers." The U.S. Department of Justice in January sued Google over similar issues. Google could use audio transcriptions or descriptions of YouTube videos as another source of text for training Gemini, leading to more-sophisticated language understanding and the ability to generate more-realistic conversational responses. It could also integrate video and audio into the model itself, giving it the multimodal capabilities many researchers believe are the next frontier in AI, according to interviews with nearly a dozen people who work on these types of machine-learning models. Google CEO Sundar Pichai told investors earlier this month that Gemini, which is still in development, is exhibiting multimodal capabilities not seen in any other model, though he didn't elaborate.
The advantage that Google gains in AI from owning YouTube may reinforce concerns among antitrust regulators about Google's power. On Wednesday, the European Commission kicked off a complaint about Google's power in the ad tech world, contending that Google favors its "own online display advertising technology services to the detriment of competing providers." The U.S. Department of Justice in January sued Google over similar issues. Google could use audio transcriptions or descriptions of YouTube videos as another source of text for training Gemini, leading to more-sophisticated language understanding and the ability to generate more-realistic conversational responses. It could also integrate video and audio into the model itself, giving it the multimodal capabilities many researchers believe are the next frontier in AI, according to interviews with nearly a dozen people who work on these types of machine-learning models. Google CEO Sundar Pichai told investors earlier this month that Gemini, which is still in development, is exhibiting multimodal capabilities not seen in any other model, though he didn't elaborate.
Oh man (Score:2)
Re: (Score:2)
Nah, they're going to train it on the comments.
I can't wait...
Re: (Score:2)
Re: (Score:1)
Don't worry. Not everyone is pewdiepie on YouTube.
Everyone acts like AI is all about video! (Score:3)
Google favors Google? (Score:2)
Noooo...you don't say.
Re: (Score:3)
Re: Google favors Google? (Score:1)
Ok, boomer.
except that (Score:2)
they better somehow filter all the spiderman/elsa crap and all the already prevalent generated random bullshit
maybe they can train an AI to flag such content
Re:except that (Score:5, Interesting)
Using natural, unscripted, spontaneous, spoken, conversational language to train models is actually a really smart move. This is the language that forms the foundation for more "academic" & formal language uses. Its development in children & adults (learning another language) always precedes academic/formal language.
My impression of the output from LLMs so far is that they've been trained on a lot of academic/formal/written language & not nearly enough on unscripted/spontaneous/spoken, making it sound overly formal & inauthentic when you try to actually converse/chat with it. I get the feeling of being "ChatGPT-splained." Let's see how this works out for Google.
Re: (Score:2)
What about copyright laws? (Score:2)
That means, even though Google/YouTube holds and shows the content, they do NOT own the content and it is not theirs to use to train AI.
We've already seen cases of this....
I signed up for YT before Google owned them and I don't recall signing over any rights to my content to them to use as they wish....
Re:What about copyright laws? (Score:5, Informative)
License to YouTube By providing Content to the Service, you grant to YouTube a worldwide, non-exclusive, royalty-free, sublicensable and transferable license to use that Content (including to reproduce, distribute, prepare derivative works, display and perform it) in connection with the Service and YouTube’s (and its successors' and Affiliates') business, including for the purpose of promoting and redistributing part or all of the Service.
You put anything on YouTube, or didn't take down content when the EULA changed, they can use it as they see fit and not pay you for it.
Re: (Score:1)
I signed up for YT before Google owned them and I don't recall signing over any rights to my content to them to use as they wish....
And yet you'll have as much luck stopping them as you'd have stopping a troll from dismembering and digesting you on a foolish adventure into an unexplored mountain cave.
Re: (Score:3)
That means, even though Google/YouTube holds and shows the content, they do NOT own the content and it is not theirs to use to train AI. [...] I signed up for YT before Google owned them and I don't recall signing over any rights to my content to them to use as they wish....
You don't need to own the content to use it to train AI. If you posted it to the public, then anyone in the public has the ability to view it and use the memory of viewing it in any way they wish. Just like an AI "viewing" it can use it to modify its model.
Re: (Score:2)
Yup. It's the outputs of the AI systems that could potentially have copyright infringement issues (most likely to occur from over-training), not the inputs.
Re: (Score:1)
You absolutely give up your rights and it's one reason why I never uploaded more than one or two videos to YouTube.
That seems like a bad idea. (Score:3)
If the chatbots start gathering comments off Youtube, say goodbye to anything positive ever coming out of them again. Holy wow. Talk about a cesspool. It makes slashdot's nazi moron and "beat up a liberal" folks look like ubermensch.
That would probably be one way to kill off this fascination with LLM chatbots for a bit though. I say, full steam ahead!
Re: (Score:1)
They've always had all the data (Score:5, Interesting)
Might backfire (Score:3)
If they admit that their AI actually CAN watch all the videos that are loaded up real-time, they'll have no excuse anymore to wait until somebody complains.
Re: (Score:2)
Every minute, 500 hours of video gets uploaded to YouTube and nobody watches them.
This AI COULD watch them, if the legislators gets to know this...
CatLand (Score:1)
You are all singing cats now. Sing, cats, sing!
wrong (Score:1)
Re: (Score:1)
I've too noticed its inaccurate but lie is the wrong word. The words or meaning it uses often sounds like the same word but different meaning. Does not seem intentional. If someone is saying something not true but they are unintentionally doing it, its just being wrong, not lying.
Re: (Score:3)
Are you a native speaker of American English, a British English or another variety of the English language? When watching English TV programmes & documentaries, I've noticed that the auto generated captioning is laughably wrong.
If Google wants to mine their uploaded videos to train their AI, they'll need to fix this first.
AI fed by AI (Score:2)
Google could use audio transcriptions or descriptions of YouTube videos as another source of text for training Gemini.
But audio transcription are generated by AI, and we were told yesterday that feeding AI with AI-generated content leads to model collapse [venturebeat.com], or in other words, reinforcement of garbage in, garbage out.
..If google can meet google (Score:2)