I am far from a China-apologist, so this criticism is nothing knee-jerk. What the US so-called intelligence community is doing to Deepseek, is a huge failure of imagination. It is obvious that state-sponsored attacks are trying to smear this brilliant AI competitor through DOS attacks, breaches, and outright legal chicanery (see Australia's new law, and the activity for one in the US, outlawing its use).
Showing that Deepseek can't provide answers to politically sensitive questions is more or less the same as boosting conspiracies and minority attacks without any fact checking (Meta, X). I'd say even more pernicious is the US model of untruth. In either case, they are similar if not the same kind of problem.
In a recent update to my notes on ML, I mentioned the following from an interview with DeepSeek's CEO:
large models made in China will be as much of a force to be reckoned with as drones and electric cars.
Indeed, the entire interview is quite eye-opening, though at the same time entirely predictable. In these three markets: drones, EVs, and LLMs, the secret sauce is doing fundamental, architectural research with confidence. That creates disruptive breakthroughs. Of course hiring geniuses who are motivated to do such is mandatory.
Advantages of Deepseek
The open source nature of Deepseek is possibly the most important advantage. Yes, there are other open source models out there, but not as efficient or as interesting.
There are several technical advantages of Deepseek which make it more efficient, and also therefore less expensive. Instead of reconsidering the many billions promised for an AI-focused build-out by Google (which Deepseek has apparently not caused any rethink regarding) and the Stargate initiative. Nothing thoughtful in these responses -- which are essentially ignoring the actual impact from the Chinese open-source AI model. And no, this is not a Sputnik moment:
While AI technology has provided hugely important tools, capable of surpassing humans in specific fields, from the solving of mathematical problems to the recognition of disease patterns, the business model depends on hype. It is the hype that drives the billion-dollar investment and buys political influence, including a seat at the presidential inauguration.
It is also an approach that seeks to advance AI less through major scientific breakthroughs than through a brute force strategy of “scaling up” – building bigger models, using larger datasets, and deploying vastly greater computational power. The disruptive quality of DeepSeek lies in questioning this approach, demonstrating that the best generative AI models can be matched with much less computational power and a lower financial burden.
The hype around DeepSeek is in part a reflection of the hype around AI. It is a reflection, too, of geopolitical tensions. Had DeepSeek been created by geeks at a US university, it would most likely have been feted but without the global tumult of the past two weeks. Beneath the panic lies fear of DeepSeek’s Chinese origins and ownership.
Yet, too great an obsession with the geopolitics of DeepSeek can distort the lessons we take from it. The promise of more open access to such vital technology becomes subsumed into a fear of its Chinese provenance. Concerns about privacy, censorship and surveillance, rightly raised by a model such as DeepSeek, can help obscure the reality that such issues bedevil all AI technology, not just that from China. Particularly at a time of threatened trade wars and threats to democracy, our capacity to navigate between the hype and the fear assumes new importance.