Setting aside the debate over the origin, details, resources, and costs associated with DeepSeek’s AI, the salient point is benchmarks indicate no meaningful improvement over competitors, including OpenAI, Gemini, and Anthropic. While the AI hype cycle demands breathless headlines celebrating a 0.9% gain in self-serving metrics, there are no signs of progression beyond the same plateau. Highly biased and censored AI is a step backward, similar to Gemini’s absurdly humiliating initial release. More on this later.
Recall the naivete of Sam Altman’s own words and the laughter it must have precipitated in the CCP politburo:
“It’s totally hopeless to compete with us on foundation models you shouldn’t try…”
The vast number of AI-generated news articles on DeepSeek prove this is pure indigenous innovation by a scrappy team of hedge fund engineers working on a shoestring budget who would never try to manipulate the stock market?! Under no circumstances would they accept illegally acquired training data from the Chinese government nor use similarly acquired hardware and IP at no cost to promote state-sponsored propaganda. None of China’s vast industrial espionage machinery in the US had anything to do with this advance, regardless of how many corporate spies the FBI arrests. The entire USIC’s alarming proclamations, warnings, and emergency remediation alerts on China’s thievery, hacking, operations, and so on, must be overreactions, and there is no proof DeepSeek was a beneficiary?
The real mushroom we see grow is the commentary from this otherwise intelligent community about the immaculate birth of DeepSeek et al which is actually from AI, not critical human thinkers. The writing is based on DeepSeek’s paper and a few benchmarks, deliberately excluding contextual realities. Their kids will be working for Huawei one day instead of FANG and they won’t understand how it happened. Don’t worry, DeepSeek, not ChatGPT, will explain it to you.
Expect a quantum surprise in the near future because the same “expert” assessments will be profoundly wrong about China, with devastating economic consequences (but they meant well). The difference is there is no propaganda incentive to make it public until it has been fully exploited for intelligence gain. When “harvest now, decrypt later” is operationalized, the only public indicator will be the accelerating transfer of wealth to China and the faster extinction of Western companies playing by the rules.
Cybersecurity and our current architecture have demonstrably failed. Incremental improvements are band-aids in a gunfight and will not help. Strong quantum-secure end-to-end-encryption will. Anything less is implicitly an exploitable back door for someone.
#quantum #cryptography #encryption? #PQC? #QKD? #quantumcomputing #cybersecurity? #nationalsecurity? #nationaleconomicsecurity? #AI
https://lnkd.in/eCeCFNux