DeepSeek: meteoric rise or deja vu?

The open-source revolution could democratize AI — or weaponize it. DeepSeek’s meteoric rise raises critical questions about trust, privacy, and the future of tech.

All the talk at the moment is about DeepSeek, and its ascent is undeniably rapid. When markets are truly disrupted, there’s always a conversation about promise versus peril.

Disruptors can be remembered as Uber’s taxis, Satoshi Nakamoto’s Bitcoin, and the birthing of OpenAI in the last few years.

Looking back at these provocateurs, society began to get up-in-arms, highly dubious while enthralled in equal measure. Eventually, when the tech became part of the furniture, the fervor died down – for a while – before cascading onto the next agitator.

DeepSeek’s blast-off to the top of the free app charts, on Apple, has woken us all with a jolt. Conversations are rife about the future of the AI market, coupled with geopolitical tensions.

Is there a parallel between Huawei and TikTok’s rise, which sparked international debate about China’s influence?

Those two companies have been dubbed part of China’s “soft power” strategy, with Huawei rolling out 5G networks to allegedly spy on foreign governments, paired with the social media giant TikTok harvesting massive amounts of user data to monitor individuals.

DeepSeek’s ascent isn’t just about innovation – it’s about global markets, privacy, and national security. The open-source approach means this powerful technology is accessible to everyone, but developers and bad actors could manipulate the software in crooked ways. Think of deepfakes, disinformation, and cyberattacks.

The question remains: should we be wary of the risks posed by this rapid rise?

DeepSeek’s rise: a geopolitical statement

The US has long been the trailblazer in AI, but China’s take on this with DeepSeek – a state-backed initiative that’s less than two years old – has been pushed with a view to toppling the US’s AI sovereignty.

Similar to Huawei’s dominance of telecommunications in the 2010s, DeepSeek’s rise represents China taking its gloves off to challenge Western ideals about data sovereignty and national security.

DeepSeek’s state-backed funding provides it with an edge over U.S. companies. With resources like 50,000 Nvidia H100 GPUs, this is no small feat – it’s a geopolitical statement, especially when you consider the sanctions and how DeepSeek is circumventing them. This sets a new precedent in the tech space

Lars Nyman, Chief Marketing Officer of CUDO, a decentralized cloud computing platform, told Cybernews.

These 50,000 GPUs – state-of-the-art graphic processing units optimized for deep learning, neural network training, and large-scale AI processing – represent a massive financial commitment, most likely from the Chinese government.

In comparison, OpenAI’s seminal GPT-3, based on Nvidia’s A100 GPUs, would have likely used anywhere between 10,000 and 20,000 GPUs.

This presents a huge leg-up for DeepSeek. It’s not just about gaining the upper hand however – it’s about blazing a defining trail in AI. In effect, we are witnessing the true dawning of the AI arms race.

Open source and its risks

The democratization of open-source AI allows a broader range of developers and companies to innovate. However, the risks here are immense.

Malicious actors may tamper with the infrastructure, reinforcing the use of deepfakes and razor-sharp cyberattacks.

This decentralized approach lacks oversight, and misuse could be rampant. Open-source collaboration could fuel data being sold on the dark web – inciting mass fraud – while an infodemic of deepfake videos could sway elections. Imagine ransomware attacks en masse, where actors impersonate individuals or businesses.

And as if China’s draconian social credit system wasn’t enough, we could edge closer to a 1984-esque reality – state-backed surveillance on a scale we’ve never seen before. Zero privacy.

Steven Hall, Chief AI Officer at ISG, a research and advisory firm, observed:

“Open-source AI models like DeepSeek democratize access to powerful technology, but they also expose vulnerabilities. There are significant privacy concerns, especially with models trained on state-controlled data, which may be used for surveillance or unauthorized data mining.”

It feels like a paradox, a Chinese AI suite democratizing the tech landscape.

Disruption or evolution

Will DeepSeek's cost-efficiency bring down prices across the AI market? Is having another major player alongside Nvidia, Microsoft, and Google healthy for the market?

We believe that large US AI companies will indeed lower their prices to remain competitive with DeepSeek as long as it remains a notable player in the market. However, trust and security will always be at the top of enterprise clients’ priorities, so I don’t see a dramatic shift in enterprise AI unless these issues are addressed.

Yaron Litwin, CMO of Canopy, privacy and data breach response software, told Cybernews:

The tradeoff, then, remains between cheap prices and data security. It’s reminiscent of fast fashion – choosing a quick, short-lasting fix at a fraction of the cost, with minimal ethics.

Apart from that, DeepSeek shows credible performance in tasks like math, coding, and reasoning – “on a par” with OpenAI’s models.

If DeepSeek also breaks through with innovations like energy efficiency, it could set future global AI standards. Should the US – namely OpenAI – respond by releasing new models, the market could either splinter or become the stage for the biggest geopolitical competition since the space race.