Unsettling Silicon Valley, the Chinese tech company, DeepSeek, unveiled R1 – a ground-breaking artificial intelligence (AI) model this week. This has led to industry heavyweights verbalising the implications of DeepSeek’s alleged achievements.
High-profile venture capitalist Marc Andreessen, for instance, deemed DeepSeek’s creation as an unprecedented breakthrough in technology. Remarkably, it appears that DeepSeek’s R1 supersedes or at least mirrors OpenAI’s O1 model on specific AI benchmarks. This raises more eyebrows considering this model cost a mere $5.6 million to train, a fraction of the expense being incurred by top-tier American firms.
The financial efficiency becomes even more spectacular in the light of US sanctions against technology exports to China. MIT Technology Review suggests DeepSeek’s triumph shows how firms are innovating for resource optimization and collaboration beyond restrictions.
CEO of Curai, Neal Khosla, presented another angle, citing the low costs as a ‘ccp state psyop’ aiming to sabotage U.S. AI competitiveness. However, his claims lack tangible evidence.
Disturbing the U.S. AI monopoly, DeepSeek might disrupt the very fabric of the U.S. equity markets, proposed journalist Hölger Zschaepitz. The low-cost, high performing model from a Chinese firm without chip access could infuse doubts over the current hefty capital expenditure in the industry.
Contrarily, Y Combinator CEO Garry Tan welcomed DeepSeek’s success as a boon for American rivals. He argued that cheaper and easier model training would boost the AI demand and ensure fuller utilisation of computation power.
Highlighting a different perspective, Meta’s Chief AI Scientist, Yann LeCun elucidated how DeepSeek’s breakthrough shows open source models are topping proprietary ones.
As the tech world debates, consumers are veering towards trying out the new product- DeepSeek’s AI assistant reigning at the top of the Apple App Store.
Original source: Read the full article on TechCrunch
