China vs. USA - The Battle for the AI Empire (or Is It Time to Roll Up Our Sleeves?)

China vs. USA - The Battle for the AI Empire (or Is It Time to Roll Up Our Sleeves?)

Which country is better in AI, China or the USA? We wrote a similar article about two years ago, where we discussed the book AI Superpowers by IT guru Kai-Fu Lee. Back then, we knew that Beijing had allocated two billion dollars for the development of an industrial park focused on artificial intelligence. Fast forward two years to 1st June, when the Beijing Academy of Artificial Intelligence (BAAI) unveiled Wu Dao 2.0 (which means enlightenment). Wu Dao 2.0 is now the largest neural network ever created and likely the most powerful. It was trained on 4.9 TB of high-quality text and image data, which is a significant amount compared to the training dataset of GPT-3 (570 GB after cleaning). It is reported that the Wu dataset consists of 1.2 TB of Chinese text data and the same amount in English, plus 2.5 TB of Chinese graphic data. In comparison, I feel like the Czech king of AI when I train networks on a few dozen GB of Czech texts. I should feel more like a rabbit. My first tentative steps in teaching AI in Czech secondary schools are hardly comparable to how China integrated AI into school curricula in 2018 and piloted the project in 40 schools right away.

China surpassed the USA in terms of the number of AI research papers published in 2020. The same report notes that American scientists' work was more frequently accepted at prestigious conferences and cited more often by researchers worldwide. In the 2021 update, I found that China now also leads in journal citations, but the USA still dominates in contributions to AI conferences.

There’s probably little point in discussing who has more data and fewer ethical constraints in handling it (China).

And what about finances? I thought that after the mega-investments from the Chinese government, China would have the upper hand over the USA. However, it seems that the USA still has the largest investment market in the world in the form of private companies engaged in artificial intelligence.

So what’s the conclusion? I still see both the USA and China as two AI superpowers rapidly advancing. Here in the Czech Republic and the EU, we write about how amazing we are at ethics and how we have talented people, but it’s clear that we are increasingly falling behind. When a leading expert in the Czech Republic and the EU applies for a grant for a Czech model, they can afford to request 3 million CZK and have about a 10% chance of being selected, with 30% of the funds needing to be covered from their own pocket. When their counterpart in Beijing applies for the same, they can ask for billions, have exponentially more data, and face fewer regulations.

I believe that to succeed, we must unite our forces in our Czech pond and stick together. We have enormous potential in smart people who can think outside the box. What if we taught our universities to monitor what is happening in the world more closely and connect them more with business? We have so many projects in our companies that we can’t keep up, and students are left doing boring work in drawers? Let’s teach our businesses and politicians what AI can do, and the whole thing will accelerate even further. Enough pondering; it’s time to roll up our sleeves!

Sources: • https://analyticsindiamag.com/china-vs-us-battle-lines-are-drawn-in-the-fight-for-ai-upperhand/ • https://blog.alphai.cz/bude-cina-za-par-let-lidrem-v-ai/ • https://towardsdatascience.com/gpt-3-scared-you-meet-wu-dao-2-0-a-monster-of-1-75-trillion-parameters-832cd83db484 • https://ceskapozice.lidovky.cz/recenze/v-umele-inteligenci-jsou-nejdulezitejsi-data-nejvic-jich-ma-cina.A190906_133856_pozice-recenze_lube? • https://blog.alphai.cz/umela-inteligence-je-v-cine-soucasti-jiz-soucasti-skolnich-osnov/ • https://aiindex.stanford.edu/report/ • https://blog.alphai.cz/co-si-precist-o-umele-inteligenci/

Originally published on Facebook — link to post