For those looking out at insights from external data, patent filings offer very specific, forward-looking clues that can indicate where a brand, or even an entire market, is headed. China has recently been leading the charge with patent filings in the AI space, led by Baidu, Alibaba and Tencent. What can we glean from the breadcrumbs they’re leaving behind?
Advancing AI in Chinese tech
It’s no secret that some of China’s largest corporations, like Alibaba, Tencent and Baidu (commonly referred to as BAT), have been moving the needle on AI innovation for years. Talks of an AI strategy in earnings calls dates back as early as 2010 for Baidu, who recruited Google Brain’s Andrew Ng in 2014. This strategy indicated the firm’s desire to publicly brand itself as a forward-thinking AI-driven organization before it was commonplace.
Brands around the world watch BAT companies closely – from local startups in China to major global competitors – as a single move into a new market, industry or technology can indicate cripling competition from companies of this size.
Intellectual property (IP) protection in the form of patents is one way to measure a company’s or a country’s innovation. According to CB Insights, taking a look at the patents these companies have filed can indicate their next steps well in advance – both in terms of geography and technology.
In late 2017, Baidu announced the launch of a new speaker, the Raven H, which runs on its conversational AI platform DuerOS and leverages tech from AI assistant startup Raven Tech, which the firm acquired the previous year.
However, Baidu began applying for international patents for a voiceprint login method using artificial intelligence, speech recognition using deep learning, and human-AI chatting methods, among others, simultaneously in Japan and Korea. This indicates potential expansion plans. Sure enough, at CES 2018, the brand announced Aladdin, a 3-in-1 smart speaker, smart lamp, and projector for the Japanese market, which was its first DuerOS-based hardware product launched outside of China.
Evaluating other indicators, Baidu announced acquisition of Japanese input method editor company Simeji in 2011, presumably to begin training its platform on Japanese syntax in preparation for entering the market.
Tencent gets in your face
As well, according to CB Insights, Tencent has been hitting the patent filings hard specifically in the area of facial recognition. In addition to 50 voice patents filed in the US, including voice processing and authentication methods presumably for its TingTing speaker, Tencent is one of the top applicants of facial recognition patents in China.
Tencent’s YouTu Lab provides image and facial recognition tech support to over 50 initiatives, like social networking platform Qzone and image processing utility software Pitu. They recently open sourced YouTu to other developers and made it accessible to users via WeChat apps, spurring concerns that this could kill smaller image recognition startups. WeChat is ultimately what gives Tencent such an advantage, enabling them to easily collect collect large amounts of external consumer data which they can use to train AI algorithms.
Some predict the brand will be using this technology to work with local governments, who will soon be issuing electronic identification cards for their citizens using WeChat or Alipay.
Not all patents are created equal
China as a whole has been outperforming the US by far when it comes to AI and deep learning patents. The country’s goal is to lead the world in artificial intelligence by 2030. QZ reports that “the number of patents with the words ‘artificial intelligence’ and ‘deep learning’ published in China has grown faster than those published in the US, particularly in 2017.”
Important to note, however, is what constitutes innovation worthy of IP protection in different nations. Depending on citations and individual country regulations, domestic patents may hold far less weight than international or triadic patents.
A triadic patent family is defined as “a set of patents filed at three of these major patent offices: the European Patent Office (EPO), the Japan Patent Office (JPO) and the United States Patent and Trademark Office (USPTO).” So although the number of Chinese patents is significant, it’s more important to evaluate the breadth of technological innovation these patents have been awarded to.
Specifically, China appears to be lagging behind in the area of machine learning. According to research by two Chinese VC firms, the US owns 32% of global machine learning patents while China owned 23% of those as of 2017.
Patents remain one of the key indicators that offer insights into where a company is focusing investment for the future. In his book Outside Insight, Meltwater CEO Jorn Lyseggen describes the value of tracking competitors’ patent applications, stating they can offer “an understanding of your competitors’ strategic purpose.”
“Patents and trademark applications are both laborious and time-consuming, and consequently expensive. A company won’t normally pursue a patent application unless it thinks it is important. Patent applications can signal new product launches or the arrival of new challengers encroaching into your area. Studying patent applications can also identify acquisition targets and, in some instances, predict acquisitions.”