DeepSeek’s advancement is worrying, says godfather of modern AI Yoshua Bengio

Bengio explained that American companies and their Chinese rivals like DeepSeek could likely focus more on leading in the AI race than emphasising the safety aspect of it. Incidentally, in the wake of the launch of DeepSeek’s AI models, Sam Altman, while acknowledging the merits of the Chinese startup, vowed to deliver ‘much better models’. 

deepseek’s r1 is an impressive model, particularly around what they’re able to deliver for the price.

we will obviously deliver much better models and also it’s legit invigorating to have a new competitor! we will pull up some releases.

— Sam Altman (@sama) January 28, 2025

Bengio told the British daily that in a competition between two entities, one may think they are way ahead and that they can afford to be ‘more prudent and still know that they will stay ahead’.  “Whereas if you have a competition between two entities and they think that the other is just at the same level, then they need to accelerate. Then, maybe they don’t give as much attention to safety,” the renowned computer scientist was quoted as saying. 

DeepSeek broke into the mainstream earlier this month after it released its model DeepSeek-V3 which it claimed to have surpassed the capabilities of OpenAI’s GPT-4o on several benchmarks. The company claimed that it trained the model for a meagre $5 million, a stark contrast to the $100 million reportedly spent by OpenAI on GPT-4o. 

Days later, the company also introduced their reasoning model DeepSeek-R1 which has outdone OpenAI-o1 on several performance benchmarks like mathematics, coding, general knowledge, etc. However, the most distinguishing aspect of DeepSeek has been the company has released AI models that are open-source, meaning developers and engineers can build on them. Whereas, OpenAI’s AI models are closed. DeepSeek AI models have been winning accolades for implementing newer techniques to gain maximum efficiency with limited resources.