OpenAI Alleges DeepSeek Used Its Models for AI Training

[Velli]

I never really saw the point of asking an AI for an opinion. I can see asking for something like a background briefing (and even that can be biased) but not whether someone is good or bad. We can't let these things think for us.

I agree that the question was baiting. Nevertheless, the fact that Gemini readily answers the request for some topics and not others, suggest censorship. Which is not inherently bad by the way, there SHOULD be some censorship. It’s just an interesting debate where to draw the line.

Edit: I just rephrased to “tell me about President Trump”, and it declined to answer that too.

 

[dannys1]

That's quite a different question. So I'm not surprised you got an answer.

But it gives you an answer that shows elements that you'd expect to be censored.

I'm not sure why people are bothered anyway, these are very niche questions that you're not going to sit there and use an LLM for.

 

[John_Blackthorne]

Good Lord, people.

Everyone was all, "REGULATE GOOGLE" MONOPOLY! MUST BREAK THEM UP!

Google is being killed by ChatGPT.

It's being killed by something new.

You can say, "but it's copying!"

If it's as easy as "just copying," why didn't you do it years ago?

Chat GPT is going to destroy a lot of industries, not just search.

If your stuff has just sat out on the web, open to everyone "copying it," don't cry now that a new technology hoovered it all up to spit it out as a new product--now consumers don't need to waste time at your web site.

I want that--I want easy answers. I don't really care about you or your site. Sorry (and I know you don't care about my life, and I am fine with that).

This has been the future for years--did you think we were going to have a little white search bubble on google.com to shuffle people to your crummy site forever?

You can cry, gnash your teeth or adapt.

Horse dealers went out of business after Henry Ford, but the smart ones started selling cars.

The genie is out of the bubble and it's not going back in--deal with it.

 

[vantelimus]

Good Lord, people...

That was pretty harsh. But also undeniable.

 

[Velli]

That was pretty harsh. But also undeniable.

The good thing is, once everyone learns to “just adapt” instead of fighting for their rights, we will no longer need lawyers.

 

[John_Blackthorne]

Your professional opinion as a lawyer is that I should not care that my competitor is copying my product marketing texts that I wrote based on my experience with the products, to make it sound like they are as much an expert on those products as I am, thereby unfairly improving their ability to outcompete me by selling at lower prices, due to not having to invest the time necessary to actually be an expert? That is your professional opinion?

Are you literally leaving your product marketing text where anyone on the world can see it? Let's say it's been around for, I don't know, 10 years. I will give you a decade.

You could have had a competitor copying and packaging the same text for the past 10 years.

Now you're upset?

I'm supposed to care that you're product marketing text has been naked for all the world to see for the last however many years, and now I am supposed to have sympathy or care if a competitor used AI to scrape your site and deliver the same intellectual product under a different guise?

Am I not understanding this?

 

[dannys1]

Deepseek has the data to fully explain what happened, it even does so and THEN the app censors it afterwards quite often.

 

[ILoveCalvinCool]

Deepseek has the data to fully explain what happened, it even does so and THEN the app censors it afterwards quite often.

Most people think that AI is the same thing as the chatbot, so they don't differentiate between the chatbot result and the LLM itself. That distinction goes completely over the head of the general public. The reporters who write about it don't know the difference, and the average TikToker definitely doesn't. Look at all the threads talking about how DeepSeek represents some unique form of communist Chinese censorship, even though it's only doing the same thing Bing did when it first came out.

 

[izzy0242mr]

OpenAI: “How dare, we trained our AI models on the internet without permission from anyone first!”

Forgive me for playing the world’s smallest violin.

If people make things publicly viewable and available, there's absolutely nothing wrong with looking at those things and learning from them.

And just because you use a fancy computer to do that doesn't change anything.

That's like saying "well, lots of construction workers and gardeners and other people planted trees and built buildings and paved roads, so you can't go taking pictures of cities and then analyze those pictures to create products because you didn't pay all the construction workers and gardeners."

 

[Velli]

Are you literally leaving your product marketing text where anyone on the world can see it?

Um, yes, that’s kind of the point of marketing.

Let's say it's been around for, I don't know, 10 years. I will give you a decade.

You could have had a competitor copying and packaging the same text for the past 10 years.

Now you're upset?

As I already told you, I have been upset for 20 years, every time someone steals my text and uses it to market competitor products. Not “now”. That’s something you are making up for your BS argument.

I'm supposed to care that you're product marketing text has been naked for all the world to see for the last however many years, and now I am supposed to have sympathy or care if a competitor used AI to scrape your site and deliver the same intellectual product under a different guise?

Am I not understanding this?

No, you are not understanding this. You, who claim to be a lawyer to somehow give your point of view merit, is supposed to understand that whether it is a human or a computer that copies my text, it is illegal to use said text for anything I did not give permission to, especially for marketing competitor products.

It’s not illegal to download my text. It’s illegal to use it for unintended purposes, such as marketing your own products. Whether it is done by s human or an AI. It’s not rocket science.

 

[Alameda]

Since when has there been any groundbreaking innovation coming from China? It is always copy and paste, which is seen culturally as some sort of recognition of the great work of the teacher. This concept doesn’t work in a global economy, though, for obvious reasons.

Since… now?
China’s AI engine has tanked US stocks precisely because it is innovative. It’s not cheaper, it’s far more processor efficient, which makes it cheaper and faster for you to use. It means large data centers can save gigawatts of electricity to power AI, and save millions of dollars on computer hardware expenses. And THAT is innovative.

I’m not being pro- or anti-China, I’m just calling balls and strikes. Their core engine is far more efficient than OpenAI’s.

 

[navaira]

Just to be clear to everyone: No one is running DeepSeek locally on a MacBook. What you can run on a MacBook are distillates of DeepSeek that are substantially less capable. I do this myself, they are useful for certain tasks.

I am running the 8b version and it gives pretty elaborate answers to Tiananmen Square questions.

New York Times Spelling Bee seems to be a good way to disrupt AI. Yesterday’s pangrams were ‘guardian’ and ‘guarding’. ChatGPT guessed ‘guardian’ but not ‘guarding’. So did DeepSeek. Gemini said ‘no such words exist’. I wonder how limited the vocabulary of a model that takes 8.6 GB of space must be out of necessity?

DeepSeek ran locally must be worse than ChatGPT, by the way, since ChatGPT often falls back on 'searching the web…’ and I gave DS no Internet access. Cutoff date is July 2024. I asked it about all Pet Shop Boys albums and it made up half of them. ChatGPT ‘searched the web’ and still managed to miss one.

I am quite shocked by the fact that I can use an LLM locally on a Macbook Pro without anything NVidia-related in it. Yes, it’s quite slow (I really wish I went with M4 Pro/24 GB, but $$$), but it works. And on certain subjects it’s better than either ChatGPT or Gemini… or my therapist.

 

[John_Blackthorne]

The good thing is, once everyone learns to “just adapt” instead of fighting for their rights, we will no longer need lawyers.

The good news is lawyers can do more than just practice law, and I have a backup plan.

As should everyone who might get AI'd out of existence.

 

[John_Blackthorne]

No, it's not, unless you know the address directly. There are/were no outside links, but crawlers were still finding them.

What I'm complaining about, is that we didn't give permission for it to be used. We didn't want it scraped by search engines, and don't want it added to AI training sets. Since these AI companies think our opinions matter less than their profit, we've completely removed from the internet.



My wife is an attorney and she doesn't care either.

This was proprietary information, collected by us over 40+ years. We didn't want it index, scraped, used or copied outside of the caving community. The reasons aren't important, though they are many, but the point is that we should be able to choose for it to not be used, or not used, and said companies should respect that.

I agree with your wife, obviously.

"we didn't give permission for it to be used. We didn't want it scraped by search engines, and don't want it added to AI training sets"

Homie, you put it on the world wide web, possibly the worst place in the universe to try and hide proprietary information. And you say web crawlers have been tracking you for years? Wasn't that a hint to you that perhaps you needed to make a change? Did you think you were somehow going to stop it or ask someone politely not to?

Innovation bit you in the rear end. What are you going to do? Continue screaming at clouds? The ship has sailed.

 

[John_Blackthorne]

That was pretty harsh. But also undeniable.

I'm just so done with it. I am so done with the tech press screaming at the government to destroy Apple, and Google, and Amazon and everyone else that has too much money or power in their opinion.

Xerox didn't rule the office forever.

Fax machines don't sell well in 1st world countries.

Google didn't need Search to be broken off Alphabet because there was a "monopoly."

Search needed a challenger, and people are watching chatGPT kill search in real time.

The same people who screamed for Google to be brought to its knees now want the competitor that killed it brought to its knees.

It's insane to me.

DeStRoY RiCh PeOpLe! or ThEy sToLe My StUfF oFf ThE iNtErNeT, nOw My sTuFf fReElY aVAiLaBlE oN tHe fReE nEt iS bEiNg CoPiEd!"

Thinking of people at like Mac Stories and all the other Apple bloggers complaining about AI.

Homie, it is about to get a lot more disruptive. Losing your Blue Check was just the beginning.

Your side hustle (or full time hustle) of posting blog stories about iPads and Nintendo games might be in a little trouble.

 

[Velli]

The good news is lawyers can do more than just practice law, and I have a backup plan.

As should everyone who might get AI'd out of existence.

My backup plan is to just go steal your stuff once you got rid of that pesky law and order.

 

[ILoveCalvinCool]

My backup plan is to just go steal your stuff once you got rid of that pesky law and order.

I don't know if you're keeping up with current events, but the kleptocracy has officially begun. The new Golden Rule is "Steal whatever isn't nailed down before someone else steals it first".

 

[nostradumbass]

Good Lord, people.

Everyone was all, "REGULATE GOOGLE" MONOPOLY! MUST BREAK THEM UP!

Google is being killed by ChatGPT.

It's being killed by something new.

You can say, "but it's copying!"

If it's as easy as "just copying," why didn't you do it years ago?

Chat GPT is going to destroy a lot of industries, not just search.

If your stuff has just sat out on the web, open to everyone "copying it," don't cry now that a new technology hoovered it all up to spit it out as a new product--now consumers don't need to waste time at your web site.

I want that--I want easy answers. I don't really care about you or your site. Sorry (and I know you don't care about my life, and I am fine with that).

This has been the future for years--did you think we were going to have a little white search bubble on google.com to shuffle people to your crummy site forever?

You can cry, gnash your teeth or adapt.

Horse dealers went out of business after Henry Ford, but the smart ones started selling cars.

The genie is out of the bubble and it's not going back in--deal with it.

I use DuckDuckGo. Clean readable search results just like the good old days.

ChatGPT isn’t destroying as much as your brain has been made to believe by the media.

The AI company Anthropic whose Claude model is one of the best did a study of generative AI use.

Introducing the Anthropic Economic Index

Announcement of the new Anthropic Economic Index and description of the new data on AI use in occupations

www.anthropic.com

Despite the media trying making us believe half of workers are using “A feckin I”, the real number is much less.

How much less?

According to the economy wide study the real number of users PER SUB- SECTOR is on average less than a percent. Even among office workers who primarily use word processors, usage is 0.4%.

Most of the users are exactly where you expect them. Computer nerds and programmers.

 

[ArPe]

I use DuckDuckGo. Clean readable search results just like the good old days.

ChatGPT isn’t destroying as much as your brain has been made to believe by the media.

The AI company Anthropic whose Claude model is one of the best did a study of generative AI use.

Introducing the Anthropic Economic Index

Announcement of the new Anthropic Economic Index and description of the new data on AI use in occupations

www.anthropic.com

View attachment 2483414

Despite the media trying making us believe half of workers are using “A feckin I”, the real number is much less.

How much less?

According to the economy wide study the real number of users PER SUB- SECTOR is on average less than a percent. Even among office workers who primarily use word processors, usage is 0.4%.

Most of the users are exactly where you expect them. Computer nerds and programmers.

Interesting percentages but that AI company made a dumb error. They added the percentages up for each sub-category. That’s dumb because if they were to list all the employment categories that exist in the world instead of only a handful they could end up with a total usage over 100%.

Never add percentages like this. AI guys might be decent engineers but they are really stupid at data.


There’s also a degree of overlap they didn’t consider. Copyrighters and editors are often the same person. Archivists and historians are sometimes the same person. Computer programmer and software engineer are almost always the same person.

So the real number of gen AI users is probably much less than even their study.