Anyone remember DeepSeek? Yes, the Hangzhou-based company that surprised other hyped and established AI startups by training its LLMs and a fraction of the cost that OpenAI was spending. After a prolonged silence, now the Chinese company says AI companies themselves to double up as whistleblowers and report job losses.
“Humans will be completely freed from work in the end, which might sound good but will actually shake society to its core,” said Chen Deli, a representative of DeepSeek at the World Internet Conference in Wuzhen. He then urged AI companies to act as “whistle-blowers” by warning the public about the jobs that would be made redundant first.
Chen was among representatives of six other companies including Unitree and BrainCo that have come to be known as “Six Little Dragons” for AI. The event, organised by the Chinese authorities, heard the DeepSeek developer go on record stating that he was quite pessimistic about AI’s future impact on humanity.
AI is dangerous but the world cannot stop its development
A report published by the South China Morning Post noted that Chen shared DeepSeek’s commitment to developing artificial general intelligence (AGI) in spite of its potentially “dangerous” impact on society. This is the second time that the company has sent a representative to an industry event.
Earlier, DeepSeek’s head of AI governance Wu Shaoqing had joined a panel on ethical guardrails around AI at the Global Open-Source Innovation Meetup organized in Hangzhou in September. This time round, Chen said AI could assist humans in the short-term but could threaten jobs in five to ten years.
AI companies need to be aware of these risks while working on enhancing the LLMs and creating new use cases, Chen noted and sounded a warning that “in the next 10-20 years, AI could take over the rest of work (humans perform) and society could face a massive challenge, so at the time tech companies need to take the role of ‘defender’.”
“I’m extremely positive about the technology but I view the impact it could have on society negatively,” he said while noting that one of DeepSeek’s strengths was its “long-term focus” while avoiding short-term trends. However, he reiterated that it was not “alarmist” to consider such systems being dangerous to society.
The future journey on AI needs guardrails
The panel noted that similar concerns were shared in an open letter sent in October where several experts sought a ban on developing superintelligent AI before a strong public buy-in was created based on a broad scientific consensus that such development could be carried out safely.
The letter was signed by several hundred AI experts, policymakers and celebrities but Chen felt that slowing down or stopping AI development would not be realistic now, given the profit incentives driving the industry. “You could even say the mark of success for this AI revolution is that it replaces the vast majority of human jobs,” he noted.
Since DeepSeek hit the headlines during January with its low-cost AI model that was seen to outperform existing ones from OpenAI and others, the company has largely remained under the radar. Barring the exception of its founder CEO Liang Wenfeng’s meeting with President Xi Jinping that was televised in February.
Not only have their leaders kept a low profile, the company itself has stayed away from the public eye, having not shared any information about updating their LLMs. However, they did reveal an upgrade of the V3 model in September. It was described as its latest “experimental” version that was easier to train and better at processing long sequences of text than the previous ones.