Can Google Challenge OpenAI With Self-Attention Patents?

The Wild West of Intellectual Property in Artificial Intelligence

In the past, intellectual property issues in AI were generally ignored. Technology has moved so quickly, most systems published in the academic literature rarely advance beyond proof of concept, industrial strength platforms have resulted in paradigm shifts only in niche markets and patent enforcement is rarely justified.

As someone who founded an AI company in 2014 on the cusp of the deep learning revolution, I still remember how it started. In 2014 and 2015, deep neural networks began to outperform humans in many tasks including image recognition and basic computer games. And while much of this research has been made public, some companies are starting to build large intellectual property portfolios. I remember it after google DeepMind acquisition We analyzed the intellectual property landscape, and many concepts published had patents pending. We decided to follow their path. At that time, in order for an artificial intelligence company to prove the valuation, he advised venture capitalists to patent some core technologies, and we had several patents granted in CopiedAnd my proteinAnd microbiome Predictors of biological age, many generative concepts in chemistry, biology and even ways to connect the two fields. Today, even a high school student might replicate this work but a decade ago, when deep learning scientists were so rare and expensive, it was risky to engage in this type of research. However, when it came to the enforcement of these patents, it became clear that it was not a common practice. In addition, it is very difficult to prove violations in server-side industrial systems. However, we have continued to patent even to prove to some partners that the software they are licensing is based on the original work. To date, we have not pursued potential infringing parties. The market for AI in drug discovery is relatively small and most of the value is not in IP around AI but in patents around specific treatment programmes.

It is likely that Google has also followed a similar train of thought when it comes to many AI startups. It was not commercially viable to sue their AI patents, and building the AI ​​ecosystem was a community effort. However, as ChatGPT, DALL-E, and other paradigm shifting tools take the market by storm, challenging Google is the dominant player in artificial intelligence, undermining its core business – searchGoogle’s attitude towards IP protection may change.

Google has a significant IP address in systems that use self-attention

And when it comes to IP protection in AI, I don’t know of any company that is stronger than Google. I remember our teams spent weeks making sure the structures were unique enough to avoid potential breaches of Google’s IP. Even DeepMind, a wholly owned subsidiary of Alphabet, has a very strong IP position in the space. basic WIPO’s search returns over 800 results.

Google is also the original inventor of the basic self-interest methods used in Transformer architectures. It was Google scientists who made fundamental breakthroughs in switch neural networks that paved the way for GPT-3. In 2017, at the Neural Information Processing System (NIPS, later renamed NeurIPS) conference, Google scientists presented a seminal paper titled “Attention is all you needBy January 2023, this paper had been cited more than 62,000 times making it one of the most cited papers on artificial intelligence.

And Weibo’s simple Google’s patent search yields a comprehensive and extensive patent covering this powerful methodology.

Apparently, the patent application was filed on June 28, 2018, which is almost a year after the original paper appeared on the prepress server on June 12, 2017 and the current state provides for the grant of the patent.

OpenAI famously avoided granting patents

OpenAI, a prominent force in the field of artificial intelligence since its inception, It was originally formed as a non-profit. I tried to find patents filed under OpenAI and found it to be a difficult task. the WIPO’s search yielded only a few results to OpenAI srl, which appears to be an unrelated company. A simple Google search returned no meaningful results, so ChatGPT asked itself.

He also correctly described the patent mode of deep learning. In fact, IBM, Google, Microsoft, Samsung, and Baidu have large IP portfolios.

And according to ChatGPT, while GPT uses self-interest, it’s not clear whether Google’s patent would cover the use of self-interest in the GPT architecture.

Will Google try to test or enforce its patents?

While Google has a very strong IP portfolio covering many areas of AI and has a patent granted on the use of attention-based sequence transfer in neural networks, it is not known for flexing its IP muscles in AI. However, as generative AI continues to sweep the market, and Microsoft integrates OpenAI tools across its vast ecosystem of its products, this situation could change quickly. And ChatGPT has a perfect answer for that.

do not be evil

In 2014, Elon Musk, one of the founders of OpenAI, became famous for making both Tesla’s patents to the public To prevent unfair competitive practices in the electric vehicle industry. OpenAI’s decision not to waste time patenting core AI technologies likely stems from this philosophy.

Although Google has pioneered key theoretical aspects of generative AI, and certainly has priority in basic neural network technologies that use self-attention, it is unlikely to take legal action against OpenAI. Patents are a great way to establish priority and protect against patent trolls but suing other scientific groups in AI is not a common practice. Especially considering the fact that OpenAI emerged from a non-profit organization run mostly by people who are dedicated to making this world a better place.