Chinese Gen AI researchers snagged more patents than everyone else combined since 2013

The World Intellectual Property Organization has counted the patents and scientific publications related to generative AI it could find between 2014 and 2023, and found 54,000 GenAI-related inventions and over 75,000 scientific publications – and that China utterly dominates the field. The Org's Patent Landscape Report – …

[JRStern]

Patents are so last millennium

At this point I thought nobody takes patents seriously. Any patent filed tends to be defensive - just proves you were there so some fool you never heard of doesn't come along five years later and claim to have invented what you've been doing all along. On their own patents tend to be incomprehensible, most are invalid if they make any sense at all because they are obvious or prior art, or are laws of nature. At best a patent comes out more like what a copyright is supposed to cover - one way to describe a process capable of infinite variations. Of course your attorney will tell you that you can claim all possible variations by using cute language. Well, go for it. Whoever gots the most lawyers, generally wins, hardly depends on what's written on the patent.

As for the papers, don't these tend to be just vanity and fluff, nobody is going to print anything VALUABLE, that's all trade secret.

Are any of these peer reviewed? Google tends to just publish whatever, they pay some academic referees but they're basically proof readers.

[heyrick]

So maybe it's not time to dismiss all that "GenAI will invent the drugs of the future, now" hype.

Seems like the business plan is to patent every molecule the AI can come up with and then if it turns out to be useful some time in the future, cash in...

[that one in the corner]

> Seems like the business plan is to patent every molecule the AI can come up with

Well, they tried damn hard to patent natural gene sequences (weirdly enough, the US actually struck that one down whilst Europe allows it[1]) so why not try that tactic as well.

Almost surprised someone hasn't just written a shell script to loop through all the elements in all combinations and try to patent the lot! Hmm, probably adding "GenAI" in the description makes the Patent Office feel that some ingenuity has gone into the process, so makes it more patentable (bleugh).

[1] at least up to 2019 AFAIK; please, please point me to something (written in English, not Lawyerese!) that says Europe has seen the light since then.

[ITMA]

The problem with patents, and this has always been THE problem is this:

A patent is only as strong as your abilitity to defend it. That, very basically, comes down to "how deep are your pockets?"

Like a lot of things in (non criminal) law, those with the deepest pockets (the most money) can ultimately win if they can keeping litigation going until the opposition run out of money.

[martinusher]

Ah, the Golden Rule -- "He who has the gold makes the rules".

[ITMA]

Yup...

Now can I have my gold sovereigns back that you borrowed?

[steviebuk]

Ironic

Considering China ignores all patients and copyright law. If they want something, they just copy it.

Remember Xi, "Don't copy that floppy".

[Anonymous Coward]

Re: Ironic

I thought the US ignores all patents as well, just relying on lawyers to prove it was ok to do that: Might is Right.

[Paul Crawford]

I wonder how many of the AI patents have been written (in part or whole) by AI?

[that one in the corner]

> I wonder how many of the AI patents have been written (in part or whole) by AI?

"We present a mathematical proof that The Singularity is logically impossible and can not be reached by using 15,000 A1000 GPUs, mwaa ha ha[1]"

[1] HAL: you may want to edit that last bit out before sending this to Nature.

[I am David Jones]

Take with a pinch of salt.

While I’m not completely sure, I think it is very likely that a “patent family” need not include a granted patent - the term is generally used without any discrimination between applications and granted patents.

And over the last 10 yrs or so the Chinese have massively increased the number of filed patent applications, flooding the “patent landscape”. And a lot of these were of a very low quality, as you might see from an academic paper mill.

So while I don’t doubt that China is a force to be reckoned with in the AI field, I am not sure that mere numbers of patent applications is by itself a reliable indicator of research activity.

[ITMA]

Re: Take with a pinch of salt.

Rough translation:

"You can file as many patent applications as you like - it doesn't mean they will be granted. Until a patent is granted, it is not a patent. A lot of Chinese applications are likely to fail as they are crap".

[Anonymous Coward]

Government influence

"Beijing has prioritized AI research for years. It was only in 2023 that US president Biden..."

Okay but you have to factor in that governments in the West have almost no influence on what industry does. Although they like to take credit when things go well. In the US, innovation is driven by investors offering huge sums of money for the best ideas, or the ability to commercialise ideas. Academics can spin off their ideas into fantastically wealthy companies. In China innovation is driven by patriotism and love of the CCP? I'm being simplistic and of course the Chinese government will throw money at research priorities but I think the US model for encouraging innovation is much more motivating.

Europe hasn't voiced any worry about Chinese companies dominating the provision of LLM access, they have said that they are worried about a tiny number of US companies dominating that space. I don't think counting patents and publications is a reflection of the real situation.

[that one in the corner]

Any citation is a good citation

> it has been cited 11,816 times. That makes the upstart the 13th-most acknowledged source of AI research.

"The practice of ignoring copyright when training LLMs, pioneered by OpenAI [Thieving Swine et al, 2019], ..."

[prh99]

Need to find a new metric

Shocker, an organization devoted to intellectual property laws likes horse racing countries by patents granted in a given field (AI being the field dejour).

Patents as metric for innovations has a lot of problem. For one it's antithetical to the concept and there is little evidence connecting the two, and when companies cease to innovate it's not uncommon to use it's patent portfolio to hold back others or worse sell them patent trolls. Or take Oracle (not exactly innovating on anything except pumping licensees for more money), one of the first thing they did after acquiring Sun was their patents to sue Google over Android and compare that to Tesla in 2015 when Musk basically said their patents were free to use.

Who knows how many of these patents will be useful going forward.