Does DeepSeek Dream of Tiananmen? Artificial Intelligence, Remembrance, and Forgetting
I neither believe in--nor welcome--the artificial intelligence singularity.
I see AI as a fantasy vision of centralized digital control promising to take as much of the world’s labor and political and social initiative as possible out of the grubby, grasping, fallible, and infallibly unionization-inclined hands of human meatsacks such as myself and turning knowledge work over to corporate owned machines curating the planet on behalf of a Mars-centric techbro elite.
Whether AI will cure cancer, lick fusion energy, or simply decimate the Philippine call center business while supplying inferior service remains to be seen.
And the idea that this supposed superpower must be held as a government-sponsored private enterprise in the hands of a consortium of patriotic tech companies is the cherry on top.
I’m guessing that to our masters the greatest perceived value of AI—may I say, the “killer app”?-- is in fighting a million-drone war with China; and the Pentagon let it be known that if the AI cabal wanted a trillion dollar payday, they better make it a proprietary, buy-America, military-industrial-complex-friendly enterprise.
So AI, at least as sold by its promoters and consumed by the US security apparatus, is regarded as the ultimate weapon, whose dominance would be ensured by the infallible method of throwing gigantic amounts of money to defense contractors on technically dubious grounds.
By their reading, American AI dominance is anchored in a US monopoly of hundreds of thousands of the best chips—the $30,000 a unit NVIDIA stuff China can’t copy..or buy. It leverages access to America’s trove of digital data—petabytes of precious training that US restrictions and Chinese censorship would deny to the PRC project. And it gobbles so much energy that it will single-handedly revive the American nuclear power industry.
The US AI adventure was capstoned by Donald Trump announcing that, in order to please his techbro backers, or at least some of them not including Elon Musk, the US would midwife the buildout and institutionalization of a quasi-national champion Stargate!!!! backed by so called OpenAI, Oracle, and Softbank and powered by $500 billion dollars of…of…public relations vaporware.
The theory was, this preemptive announcement of an invincible combination of American money, will, and knowhow would obtain the capitulation of the world’s users and erase competitors a.k.a. the People’s Republic of China from the world technological map.
Under the guidance of infallible China-war Great Helmsman Jake Sullivan, the United States had already made the decision to weaponize AI as a foreign policy tool and a benchmark for the success of his China-containment policy. Allies and fencesitters would have to toe the American line if they wanted to access the US superweapon!
In the Middle East, as part of Sullivan’s strategy to spade out PRC influence, Saudi Arabia and the UAE and who knows who else were put on notice that a condition of access to US tech and NVIDIA chips was divestment from Chinese AI ventures.
Well, a funny thing happened on America’s road to 21st century hegemony.
A PRC operation called DeepSeek dropped a nifty little artificial intelligence program that performed pretty much as well as the expensive versions that OpenAI had dribbled out.
