Let’s Accept That AI Leadership Is Everywhere
Artificial intelligence (AI) is the new space race, or at least that’s how it’s being perceived in the general population, and being presented by the movers and shakers of this world.
That’s why it was no surprise recently when the American CEO of a US-based company raised an alarm by declaring, according to the Washington Post, that “China has taken the lead on the AI revolution.” This is consistent with my observation in a Datanami column last summer that many business leaders are applying an outmoded Cold War mentality to the thoroughly transnational reality of AI as a force in 21st century global economic development.
To bolster this assertion, the CEO, in front of his counterparts in Davos, offered personal anecdotes about China’s growing AI capabilities. He declared that he’s seen an explosion of new AI businesses there, as evidenced by an “endless stream of people who are showing up developing new companies. He also noted that the “venture business there in AI-oriented companies is really exploding with growth.”
However, these assertions are not corroborated by primary market research, which shows a fairly level playing field among the world’s largest economies in AI-related investment, research, innovation, and adoption. For example, consider the following findings from Stanford University’s most recent “AI Index” study, which surveyed AI activities by academics, vendors, enterprises, and others around the world:
- AI is now the dominant focus in computer science everywhere, as measured by the growing volume of published research studies.
- Everybody is using the same open-source AI tools, especially TensorFlow, PyTorch, and MXNet. There is nobody in any country these days who hasn’t built their AI practice partially or entirely on stacks incorporating these and other AI frameworks, as well as Hadoop, Spark, and other key open-source codebases.
- Most AI R&D activity everywhere is focused on computer vision, pattern recognition, robotics, and natural language processing. This common research focus is enabled by the fact that open-source communities have accelerated the cross-border sharing of data, libraries, models, research findings, and expertise.
- AI startup activity is increasing exponentially year after year everywhere, faster than startup activity in other sectors.
- AI VC funding is increasing exponentially year after year everywhere, faster than such funding in other sectors.
- AI job openings are increasing exponentially everywhere.
- Undergraduates everywhere are flocking to AI majors and curricula in growing numbers.
- Most new AI R&D is in machine learning and neural networks, which have become the core approaches everywhere.
- Various AI capabilities are being adopted relatively equally by users across regions of the world.
- AI is mentioned most in earnings calls in specific industries, with the information technology, consumer, financial, and healthcare sectors most prominent in reporting strategic applications of the technology.
- The absolute output of AI R&D research studies continues to grow significantly everywhere.
- Most AI R&D everywhere is from academic institutes.
- AI patent activity is growing exponentially all over the world.
This level playing field is due in great part to the radical cross-border openness that has characterized the AI revolution in recent years. Even when clear national differences in AI activities reveal themselves in the Stanford study, they don’t play into a tidy us-versus-them narrative with regard to the US and China:
- Europe generates far more published AI R&D research studies than China, the US, and other regions. One might speculate that this is because European-based researchers have fewer commercial opportunities than their peers elsewhere to cash in on their discoveries, so they tend to overcompensate by publishing more work. Or, on a positive note, it may be due to the fact that democratic nations are often magnets for the smartest people from everywhere to study and do their best work.
- US and Europe have far more AI R&D coming from the private sector and China far more from government R&D institutes. This might be due to the fact that the private sectors are larger shares of US and European economies than in the People’s Republic of China, which is still nominally Communist and very much a state-dominated society.
- US AI R&D researchers’ papers are cited more by their global counterparts (in China, Europe, and elsewhere) than otherwise. One possible reason for this might be that more innovative AI research is coming from US-based researchers—many of which, such as the deep bench of AI researchers at Google, are based in Silicon Valley. Note that many of the researchers who contributed to the Beijing Innovation Center for Future Chips’s recent “White Paper on AI Chip Technologies” are in fact based in US research institutes. The fact that every one of those researchers has a Chinese name just shows how global the AI R&D community truly is.
- AI patent activity is far greater in the US, South Korea, and Japan than elsewhere. This might be due to the fact that patent activity tends to be correlated with the vibrancy of those market economies, compared with the more centrally planned economies of Europe and China.
Let’s nip this nationalist mentality in the bud. Let’s accept that AI leadership will remain a transient geopolitical asset in a world where every nation will continue to:
- Struggle to distinguish “AI output” with strategic military potential from dual-use AI technologies that are profitably applied to domestic and overseas opportunities of a commercial nature;
- Enjoy free and open access to all the tools, data, expertise, and research needed to build high-quality AI for every conceivable application; and
- Leverage the latest and greatest GPUs, TPUs, ASICs, FPGAs, systems on a chip, and other chipsets available for training and inferencing of all of this fancy AI in every conceivable deployment and application scenario.
In his recent study “Reframing Superintelligence: Comprehensive AI Services as General Intelligence,” K. Eric Drexler hit the nail on the head when he declared that “current trends suggest that superintelligent-level AI capabilities will emerge from a distributed, increasingly automated process of AI research and development, and it is difficult to envision a scenario in which a predominant portion of overall AI capacity would (or could) emerge and be confined in ‘a box.’” This observation applies just as much to the transnational, decentralized, multi-organizational AI R&D community as it does to the increasingly distributed AI DevOps toolchain that sprawls across a diversified multinational enterprise.
If China or any other country were to become a predominant AI power, the proof would be in whether they’re dominating global markets with products and services whose primary differentiation is grounded in machine learning, deep learning, and the like.
It’s not enough to assert that China’s or any other country’s AI dominance is predicated on their level of investment in one specific application, such as facial recognition or self-driving vehicles. Everybody has or will have access to those technologies at approximately equivalent level of sophistication. Even AI-powered weapon systems of equivalent prowess will almost certainly spring up everywhere at approximately the same time, in spite of export controls on the underlying technologies.
Clearly, many nations have experienced an explosion in AI funding, startups, patents, research, hiring, degrees, product, and so on. But none of these metrics can be used to definitively declare anybody’s “leadership” in this multifaceted arena.
About the author: James Kobielus is SiliconANGLE Wikibon‘s lead analyst for Data Science, Deep Learning, and Application Development.