This article was written by a young college Republican in South Carolina.
Forgive my knee-jerk reaction but there is simply no shot that China developed DeepSeek in the manner and with the budget that they claim (keyword claim). You’re telling me that DeepSeek was developed for no more than six million dollars and uses the less advanced H800 Nvidia microchips? Well I have an AI startup in Brooklyn to sell you. But wait! Before you sign on the dotted line for your new AI startup, let’s critically examine the external factors at play with DeepSeek that just don’t jive with its sudden emergence as the new ChatGPT.
Two relatively minor points. We know that the CCP has several intelligence and theft operations aimed at bringing advanced microchips and blueprints thereof. Take a gander at just a few examples from the past decade here, here, here, here, and here. These operations are significant and widespread, and only account for those that we’ve caught. The other reality that’s hard to ignore here: China has 1.4 billion people, and thus has an absolutely enormous talent pool of possible coders. They likely have millions of competent coders, and they’re getting better as China continues to siphon off effective teaching pedagogy and advanced curricula from US schools and universities. Frankly, if China can only muster an on-par AI model to challenge popular US models with so many coders—and terabytes of stolen intellectual property, mind you—that’s an ominous sign for their future as a global superpower in the AI Century.
Most importantly in the quest for truth about DeepSeek is the Taiwan-shaped elephant in the room. The CCP still sees Taiwan as a must-have possession for more than ideological reasons. The semiconductor manufacturer TSMC is Taiwan’s (and the world’s) foremost source for advanced microchips. UMC, another semiconductor manufacturer in Taiwan, also holds significant market share. The gap between these two Taiwanese powerhouses and the field is huge. Why, if China can so easily and so cheaply create a competitive AI, does it want Taiwan?
A conflict over Taiwan is a risky proposition, and could easily backfire if the US gets involved. Yet China is slowly creeping towards it. Are the many recent military exercises in the Taiwan strait just a charade? Don’t forget that TSMC has a not-so-secret contingency plan to incapacitate their own facilities if an incursion does take place, of which China certainly knows about. China apparently sees these potential bad outcomes as tolerable risks thus far, though that may change.
Don’t mistake this author’s skepticism on probable propaganda from the CCP for a dismissal of China’s AI capabilities in the future. DeepSeek is impressive in its power, and is in some ways comparable to ChatGPT. However, it was very likely developed with mountains of stolen IP, countless coders, a titanic budget, and smuggled advanced microchips from the United States. Even with these probabilities, we should watch the Red Dragon to see what it does next.
The post In Brief: Of Course the Commies Didn’t Eat Our Lunch on AI appeared first on Palmetto State Watch Foundation.