Get all your news in one place.
100’s of premium titles.
One app.
Start reading
Crikey
Crikey
Comment
Wanning Sun

What does the man behind DeepSeek want?

There has been wall-to-wall media coverage in Australia of the launch of Chinese AI model DeepSeek. Reactions have been either hysterical about the risks it poses to national security, or hyperbolic about its potentially devastating impact on the US AI industry. Descriptions of its launch as a “Sputnik moment” causing “carnage” in the tech market have provoked visceral reactions of both fear and excitement.

Many have observed the apparent demure demeanour of the man behind DeepSeek, and there have been frantic attempts to make sense of the geopolitical implications of the model’s unexpected success. But mostly unremarked so far is the potentially revolutionary impact of DeepSeek’s business model on China’s own science and technology sector.

The nerdy, media-shy 40-year-old Liang Wenfeng — now a national hero in China, and flavour of the month across the globe — remains low-key. So far, he has given only two interviews about DeepSeek, in 2023, both of which were to an obscure Chinese blog. In these interviews, Liang displays none of the bravado expected of a successful entrepreneur.

For now, at least, he’s chosen to stay in China instead of going to the US, and none of his team members have studied overseas. But in listening to his interviews, you are unlikely to get the impression that he wants to be the poster child for China’s nation-building project. Nor does he sound as if he’s in a rush to make a quick buck — although perhaps that’s understandable: he’s already become a billionaire, and his company High-Flyer has been using AI to predict patterns in the stock market.

Instead, he seems animated by innovation and an intention to shake up China’s AI industry leaders by forcing them to abandon their current mindset. In fact, he’s effectively telling China’s AI players to either throw away their playbooks or move over for the new generations of younger, bolder talents who are daring to think outside the box.

When asked why he’s not interested in developing both AI models and applications, as most other Chinese companies do, instead focusing solely on research and exploration, Liang says his ambitions lie elsewhere. He wants to join the global wave of innovation as a participant, not as a follower. This may sound trite, but in the current state of play, this is tantamount to issuing a challenge to his peers in the field:

Innovation such as ours happens all the time in the US. The Americans are surprised by us now mainly because we’re a Chinese company, and we are entering their game as an innovator with an original contribution, not as followers.

Liang made it clear that his ultimate goal is to become a leader, not just in AI but in AGI (artificial general intelligence). To do this, unfettered exploration is essential:

For many years, Chinese companies have mostly been used to letting others innovate and then monetising their innovations. But that doesn’t have to be the case. Our point of departure in this wave is not to make a quick buck, but to get to the forefront of the technology so we can drive the development of the entire ecosystem.

Liang warns that unless this mindset changes, China will always be a follower. In his view, Chinese tech companies have been held back by the inertia of the past. What the Chinese tech sector lacks is not capital but confidence in its capacity to drive innovation:

From investors to big companies, everyone [in China] felt the gap was too large, and suggested sticking to applications. But innovation requires confidence first. … We often say there’s a one- or two-year gap between AI in China and the US. But the real gap is the difference between imitation and originality.

Liang does not seem overly concerned about future challenges to DeepSeek from competitors:

In the face of disruptive technology, any moat created by closed source AI will be short-lived. Even though OpenAI is closed-source, this won’t stop others from catching up. So, we’re embedding value in our team. Our colleagues grow in this process, accumulate a lot of know-how, and build an innovative culture in the company.

Liang reminds people that innovation is not just about making money; it is more about pursuing their curiosity and a desire to create. That’s perhaps why, when asked how DeepSeek will try to protect its “moat”, Liang seems quite sanguine and almost insouciant:

By open-sourcing our code and publishing papers, we don’t lose anything. For tech people, being followed is a real sense of achievement. Open-sourcing is more of a cultural behaviour than a business one. Giving away is actually an additional honour. A company doing this also creates cultural appeal.

Son of an elementary school teacher, Liang grew up in a small town in Guangdong in the 1980s. He believes the key to cutting-edge research is a capacity to recruit high-calibre researchers from the next generation who dare to be idealists. Always on the outlook for young people who display curiosity and imagination, he has built his team from graduates and students from top Chinese universities to staff his research team.

According to Liang, this kind of confidence is usually more apparent in young people. These developers belong to a generation of young Chinese who want to enter “the game” not primarily driven by money — although he does pay them extremely competitive salaries — but by a creative urge and a broader commitment to be a global innovation leader. This, in a Chinese tradition that mostly values job security, seniority, experience and academic credentials, is nothing short of revolutionary.

Liang readily admits that the top 50 talents in AI may not be in China. But he is confident that his team definitely includes many of the top 50 talents in China.

It is not clear how long Liang can afford to maintain his idealism. But perhaps the US tech companies should be less concerned about his new AI model and more about a new generation of Chinese AI researchers. Based on the promise shown by DeepSeek, they may just have the drive and intellectual capacity to beat the Western AI establishment at its own game.

Have something to say about this article? Write to us at letters@crikey.com.au. Please include your full name to be considered for publication in Crikey’s Your Say. We reserve the right to edit for length and clarity.

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.