Is DeepSeek an AI breakthrough or just another incremental step?

In the midst of widespread excitement about DeepSeek's implications for international AI advancement, some experts caution against overemphasizing its immediate influence. We discussed China's release of DeepSeek-R1 with Bassam Aoun of DocketAI, which is recognized for pioneering one of the world's first AI-driven sales engineers.

A Steady Progression: "Recent conversations about AI sometimes suggest innovation has slowed, noting periods without major announcements from industry leaders like OpenAI and Anthropic," Aoun remarks. "However, innovation doesn’t always occur uniformly. Progress frequently comes in bursts or exponential leaps."

While many view DeepSeek-R1 with enthusiasm, Aoun describes it as an expected milestone within an ongoing innovation process. "It’s a meaningful step forward, but one among many," says Aoun. "AI has been evolving steadily for years, and there remains considerable progress ahead before AI achieves widespread adoption and impact."

Rapid Generational Cycles: Aoun highlights the progressively brief intervals between new generations of AI models, making direct comparisons across generations challenging. "New AI model generations now typically emerge every seven to eight months," he explains, "so if a new model like DeepSeek appears after several months and surpasses previous models, this isn't unexpected—it simply belongs to a newer generation. In this instance, DeepSeek-R1 notably achieved performance comparable to earlier models but at significantly lower costs, reflecting expected incremental improvements."

China’s accelerating investments: Following DeepSeek-R1's launch, China's commitment to advancing AI is expected to intensify further. "China is likely to significantly increase its investment and resources toward developments like DeepSeek-R1," Aoun acknowledges, highlighting the growing global efforts in AI innovation.