The 2018 AI Index Report, developed by scientists and researchers in the AI field was recently released. This report uses quantitative data, such as publication counts or mentions of AI, to assess the state of AI today.

I've said a couple of times that everything in 2019 seems to be 5G, AI, or autonomous vehicles. Cadence, of course, works with customers in all three areas. But AI (and machine learning, which is a term I use pretty much interchangeably) is not just something on which we work with customers. We use it ourselves too. I think that there are two big trends in the EDA big picture. They are cloud computing, by which I mean updating our algorithms to take advantage of effectively unlimited numbers of servers. And artificial intelligence, by which I mean using machine learning approaches to automate iteration during the design process. A lot of design consists of "run a tool, look at the result, tweak some parameters, run it again." Deep learning offers the possibility of making that process happen automatically, the trendy phrase being "no human in the loop."

We've also been talking about it a lot on earnings calls. According to the report, for mentions of machine learning, the list of companies (2017 numbers) is:

• Alphabet (Google) with 57 mentions
• NASDAQ with 26
• ServiceNow and Progress Software with 21
• Cadence with 20

It may not be an official government policy, but US industry and academia have clearly prioritized AI. I recently came across the phrase the G-MAFIA in this context, which makes it sound suitably evil, but is just the initials of 6 companies. They are Google, Microsoft, Amazon, Facebook, IBM, and Apple. All of them have huge AI efforts. That's not to say nobody else does. In some ways, all of tech has prioritized AI, and the G-MAFIA are just the biggest tech companies, and so the biggest investors. But despite the G-MAFIA, it is very international. In parallel with the G-MAFIA are BAT in China, which are Baidu, Alibaba, and Tencent, also putting a lot of effort and investment into AI.

According to the report, 83% of the papers on Scopus are published outside the US. Based on the G-MAFIA and BAT, you might expect the US and China to lead the publication numbers, but actually, Europe does (see the graph above).

This fits my prejudice that Europe is good at setting up universities and, especially, research institutes that do publishable research. But they have a hard time of creating businesses that capitalize on the work. I believe that the best way to do technology transfer is inside people's heads, but if everyone is too risk-averse to leave their safe research institutions then the companies are never founded. My experience is mostly from living in France for several years. There, another problem is the government's love of French champions. I remember asking someone in Bull why there were no Sun Microsystems in France. Bull was the sort of French national champion in computers, like ICL in the UK, Siemens-Nixdorf in Germany, or Olivetti in Italy. He replied that if someone managed to get a French equivalent of Sun started, the government would dump money into Bull to enter the workstation market. They would be successful enough to overwhelm the upstart French company in France, since other French champions would be forced to buy from them, but not successful enough to be a success in other countries in Europe, let alone globally.

Here's an amazing statistic I came across a couple of years ago in The Economist. The youngest company in the CAC 40  was founded in 1967, over 50 years ago. The CAC 40 (the "cack karant") is the French equivalent of the Dow-Jones Index. The average age of companies in the CAC 40 is 101 years, whereas the average age of a company in the S&P 500 has decreased from 61 years in 1958 to 18 years now.

As to the topics of those published papers, machine learning, neural networks, and computer vision lead the way. To be honest, I think this is just a phenomenon from searching for specific words. Almost all modern computer vision involves some sort of neural networks, and training those networks involves some sort of machine learning. Not everything is vision, of course—Alexa requires you to talk to her—but there is a huge overlap between the topics.

Machine Learning and Neural Networks (followed by computer vision) subtopics lead in AI papers published

It's not just the big guys though. The chart above shows the cumulative number of AI startups. One takeaway is the vindication of the throwaway line that an AI startup is any startup since about 2014...whatever they are really doing, they will claim to be an AI startup to get funded. Nevertheless, the number of active venture-funded AI startups in the US has more than doubled since 2015.

One big message that comes across from reading through the report is just how global AI is:

Still, we can assert that AI is global. 83 percent of 2017 AI papers on Scopus originate outside the U.S. 28 percent of these papers originate in Europe—the largest percentage of any region. University course enrollment in artificial intelligence (AI) and machine learning (ML) is increasing all over the world, most notably at Tsinghua in China, whose combined AI + ML 2017 course enrollment was 16x that of 2010. And there is progress beyond just the United States, China, and Europe. South Korea and Japan were the 2nd and 3rd largest producers of AI patents in 2014, after the U.S. Additionally, South Africa hosted the second Deep Learning Indaba conference, one of the world’s largest ML teaching events, which drew over 500 participants from 20+ African countries.

Learn More

Read the AI Index 2018 Annual Report.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.