New Tech, Old State: Are we ready for an AI Revolution?'
Read more
DeepSeek: How little we understand AI
The rapid rise of DeepSeek, the Chinese AI app that has disrupted the global technology market, has been nothing short of extraordinary.
Over the weekend, DeepSeek became the most downloaded free app on Apple's US App Store. By Monday, its ascent had triggered a significant sell-off in major tech stocks, prompting concerns about the United States’ position as the leader in AI development. OpenAI’s Chris Lehane described it as "AI's Sputnik moment", drawing parallels to the Soviet Union's launch of the satellite which ignited the space race.
Developments in AI continue to confound observers, highlighting how little the world truly understands AI and its implications. Levels of excitement and the sense of opportunity are high, but amid profound uncertainty. The Public Digital opinion: AI has real potential to be transformative, but it is essential to tread with care.
Some doubt whether DeepSeek’s creators truly built the model with as little time and money as they claim. However, it does seem that this AI – comparable to US tech giants’ billion-dollar models – has been trained for less than $10 million. While Open source AI models and more efficient training strategies have been demonstrating increasingly strong performance for a while, DeepSeek has smashed all expectations.
The result? It's been a chastening few days for data centre builders, chip manufacturers and the Silicon Valley luminaries who were confident they had a commanding lead in AI. Stock valuations will no doubt continue to bounce up and down, highlighting the importance of matching ambition with a healthy dose of caution when new technology is being explored. This DeepSeek-driven volatility reminds us how little we know about this technology. How it will scale in performance; what approaches to building it will succeed; and who will emerge as leaders in its development and commercialisation.
As Public Digital’s Dave Rogers noted in his talk in October, we simply don’t know what AI will turn out to be. It is a mysterious, opaque technology that is not even fully understood within its own field. It is genuinely novel: its properties are emergent. There is no better time to reflect on Dave’s point that we should be wary of false certainty when it’s presented to us, especially in the form of grand claims about the future.
Organisations rushing to adopt AI risk unintended consequences – amplifying biases, creating new dependencies on opaque systems or creating long term costs. Rather than chasing the hype, leaders should invest in the foundations for success with AI: addressing the basics, prioritising people, adopting flexible models and building their capacity.
Whether it’s the promise of efficiency savings that will be made, or problems that will be solved, it’s easy to buy the dream. After all, the fear of missing out is powerful. There are always people willing to sell us a vision of the future, but bear in mind: they don’t know what’s going to happen either.
Written by
Peter Chamberlin
Senior Director