The impact of DeepSeek on AI innovation cycles might be less than we think

While talk is hot about DeepSeek and what it means for U.S.-China relations (and AI’s trajectory as a whole) some skeptics question its broader impact. We spoke with Bassam Aoun, Vice President, AI/ML of DocketAI, the developer one of the world's first AI sales engineers, to discuss what’s revolutionary (and not) about China’s DeepSeek. 

A drop in the bucket: "There has been some talk recently about AI plateauing, why OpenAI and Anthropic haven't released anything," says Aoun. "Innovation doesn't always happen weekly. Sometimes it happens in a stepwise function or exponentially."

While many brim with excitement for DeepSeek, Aoun sees this progress as a more natural part of the innovation cycle. "It’s notable, but not revolutionary. It’s a drop in the bucket overall. We’ve been on this march for many years, decades, and we still have a lot to go," Aoun says, adding his own ideas about what’s to come, "It’s a very long way for AI to make its impact or to become ubiquitous."

Short generations: Aoun notes that AI development cycles move quickly, making direct comparisons difficult. "Time generations [for model releases] in AI are very short," Aoun points out, while discussing the close windows between previous uodates from Anthropic and OpenAI. "We expect a certain percentage of increased efficiency every seven to eight months because we have spent enough time engineering fine-tuning cycles. [DeepSeek] has the same quality at a much lower cost, so it’s in a different generation in terms of time."

China: China’s investment in AI is only set to accelerate following DeepSeek’s debut. "The Chinese government is going to throw money at DeepSeek," says Aoun. But he emphasizes that the U.S. remains the dominant force in AI development. "The USA has a much better training budget, bigger access to infrastructure, and has built millions of high-end AI chips. We have a longer runway and more resources in terms of AI infrastructure technology."