A Lesson from DeepSeek to HR in the Age of AI

R1, a powerful new open-source artificial intelligence (AI) model created by Chinese startup DeepSeek, has sparked praise, concern, and questions about who is now leading the AI arms race. Developed by relatively junior engineers at a fraction of the cost of similar models, R1 is rivaling OpenAI’s o1, indicating a major transitional moment in the tech space. 

Personally, I’m convinced that this is a game-changer that nobody saw coming, and from it, we can gather a few things: trying to safeguard resources and know-how clearly leads to the opposite results—instead of thwarting, it spurs innovation. The mad dash to more resources, money, and computing power is a bubble. The same or better results may be achieved with much less. Finally, the desire for AI dominance will now go into overdrive. 

That said, and despite the initial knee-jerk stock market reaction, this isn’t necessarily bad news. Rather, it’s a much-needed shift in perspective that will democratize AI use significantly. The open-source nature of DeepSeek has huge advantages for enterprise use. Having less of a black-box solution, more flexibility for organizations to optimize Large Language Models (LLMs) for their specific needs, and the ability to train and run said LLMs on their own hardware and private cloud are huge positives. 

The latter will vastly improve data security and will further remediate privacy concerns. Despite TikTok-induced cold feet about data sharing with Chinese servers, this should mitigate concerns around sharing it with the likes of DeepSeek (the chatbot) and, domestically, OpenAI and Microsoft. 

Here, you have an opportunity to deploy the model in your private cloud. HuggingFace is among the first companies to offer hosting in the US. Perhaps the most obvious is the cost. Cheaper services and free software paired with much lower computing requirements will further reduce the entry barrier.

So, what does this have to do with human resources (HR)? Despite being a field not often lauded for its technological savvy, it is actually a lot. Every organization hires. Most use some form of AI in their business operations, meaning the need to fill technological roles has increased. According to research from the Bureau of Labor Statistics, computer and IT occupations are expected to grow much faster than average from 2023 to 2033, with a projected 356,700 job openings annually.

The problem? Tech is moving faster than the pace it’s taking talent to achieve the hard skills desired by employers to fit these roles. By 2026, the IT skills shortage will become a significant problem for most organizations globally, resulting in $5.5 trillion in losses, according to an IDC survey. And the impulse to dump more and more money into this challenge is clearly not the answer. It’s not even about bringing in more experience. It’s about changing the way we hire entirely. 

Of the R1 news, The Wall Street Journal captured a sentiment that will (or at least should) redefine HR in the coming years: “Take a team of young Chinese engineers, hired by a boss with disdain for experience. Add some clever programming shortcuts, and a loophole in American rules that allowed them to get advanced chips.” It wasn’t deep pockets or tech heavyweights that launched a relatively unknown startup into the arena with OpenAI. It was a new approach applied to an old problem. 

Experience has been the key metric for success as far as hiring goes. However, with AI taking on more and more technical tasks, hard skills and time spent in an industry are soon going to be overturned by “grayer area” factors, like potential and soft skills. These capabilities include strong communication, critical thinking, and the ability to adapt to change and work well with others. Frankly, it’s a welcome and much-needed shift that will level the playing field for job seekers and the future of (human) AI talent. 

Take prompt engineering, for example. The idea of this being an essential skill to apply to AI has petered out. Reasoning engines do a good job refining prompts, and people no longer need to be particularly crafty with their initial asks. Now, if you’re working with AI, it’s far more important to express what you need in the most basic and clear way. This is much more of a soft skill, and identifying that all comes back to a company’s ability to hire the right talent.

You need people who can accomplish things that haven’t been done before. In the case of DeepSeek, engineers thought creatively within the constraints of a small budget and the sanctions put forth by the US government. This should be a wake-up call for HR to reevaluate how they determine what qualifies as top talent and how they can achieve more with less. 

If that’s not enough, let’s look at bigger economic trends. Nearly half (41%) of employers expect to downsize their workforces with AI, according to TechBrew. As AI systems grow more autonomous, many hard-skills-based functions can be replaced by chatbots. People would rather hire robots than recent college grads. But AI can’t perform many of the soft skills that make us uniquely human. And if we want to keep employing them, that is the basis HR should be evaluating them on. 

Despite being an exciting time for AI (and as a co-founder of a company in this space), much of it has been overhyped. We always talk about the next big thing that will change everything. Meanwhile, the actual deliverables are often underwhelming. But as we’ve seen by the headlines, innovation doesn’t always come from the biggest players. However, it comes from the right talent under the right circumstances. And the companies and HR teams that hire smartly will be the ones that gain the edge in the race to AI supremacy.