You’ve seen the headlines, but what is DeepSeek, and—more importantly—is this something HR leaders need to care about?
DeepSeek, a two-year-old Chinese startup, launched its latest model early last week, setting new benchmarks in AI performance with top-tier models from Anthropic and OpenAI. According to a LinkedIn post by Armen Berjikly, chief technology officer at BetterUp, DeepSeek “found a series of clever engineering solutions that reduced expensive memory usage with minimal impact on performance.”
These breakthroughs have sparked a conversation about global AI leadership and the sustainability of tech valuations. The global investment strategy team at J.P. Morgan Wealth Management described the situation as “markets adjusting to a new threat,” calling DeepSeek a model that challenges “assumptions about U.S. technology dominance.”
DeepSeek’s latest model achieves performance on par with OpenAI’s cutting-edge model in math, coding and reasoning tasks, according to J.P. Morgan Wealth Management, while reportedly operating and training at approximately 95% lower costs.
Berjikly explained that the model is cheaper to operate than traditional alternatives, owing to advancements like multiple-token prediction, which speeds up response generation with minimal accuracy trade-offs.
Behind the scenes, DeepSeek activates specific parts of the model for specialized tasks, reducing unnecessary memory usage, notes Berjikly. Additionally, it uses 8-bit precision instead of the standard 32-bit for training, significantly cutting hardware requirements. Berjikly, who founded Kanjoya in 2005, which UKG later acquired, further explained that DeepSeek employs reinforcement learning without human fine-tuning, enabling robust reasoning capabilities.
Joel Sadhanand, director of talent development at Comcast, envisions an HR department where hiring is faster, employee engagement is stronger, and strategic decisions are guided by data rather than intuition. In his recent LinkedIn newsletter, Sadhanand asserts that with DeepSeek, this vision is no longer a distant dream but a present-day reality.
While more hands-on research will come, Sadhanand believes DeepSeek already is transforming HR by streamlining hiring, onboarding and workforce planning with predictive analytics and automation. He says the tool also has the potential to enhance engagement, retention and performance through real-time feedback and sentiment analysis while supporting employee growth with tailored training and well-being initiatives.
HR tech in the news
Global remote employment servicer Remote has launched Contractor of Record, a platform that streamlines hiring and paying international contractors by managing compliance, administration and payments.
Performance management platform 15Five has acquired Kona, an AI-powered assistant for remote managers. This integration will enable 15Five customers to scale manager support, addressing performance, engagement and retention challenges.
Veritone analyzed 2024 U.S. labor data, revealing that AI outpaced the broader job market with 81,298 positions added in 2024—a 24.5% increase from 2023. AI job openings rose 36.6% year-over-year, with median salaries reaching $151,996, up 2.7% annually.
More from HR Executive
As HR leaders plan talent pipelines, many are exploring new global recruiting pools. Could Switzerland be an untapped treasure for recruiters? Experts suggest this country could rival the San Francisco Bay Area as the world’s leading deep tech talent epicenter.
Josh Bersin describes today’s EX challenges as a “soup” of issues that fall squarely on HR’s plate. To meet these challenges, HR leaders must focus on simplifying work processes, enhancing flexibility and rebuilding trust, according to experts.
Do SMBs have an AI advantage? Small and medium-sized businesses (SMBs) could be the world’s trailblazers in the practical adoption of AI, says Guillaume Roy, chief innovation officer and co-founder at employee experience platform Workleap.
Credit: Source link
