“Lost in Translation: China’s ChatGPT Launch Stumbles in its Debut!”
China’s version of GPT-3, the natural language processing artificial intelligence model, stumbled with its first lines. The Chinese AI model, called Wu Dao, was introduced by the Alibaba research team as their flagship product. However, the chatbot failed to impress at its launch, struggling to maintain a coherent conversation.
Unlike GPT-3 that has access to data from the English language, Wu Dao was trained on a large-scale Chinese language dataset. Wu Dao can comprehend 1.75 terabytes of text, which covers over 1.3 million books and documents, compared to GPT-3 which can hold up to 570GB.
Despite its impressive size, WuDao’s responses were not persuasive, persuasive, or natural-sounding. During a public demonstration, Wu Dao had difficulty answering questions such as “What can you do?” and “What do you think of humans?” Out of curiosity, a person asked the AI model if it could recognize that it was its birthday, but the chatbot couldn’t do so.
The Alibaba team stated that Wu Dao is still in its development phase and that they plan to fine-tune the technology to make sure it operates more effectively. For now, though, the technology only appears to be of interest for tech enthusiasts and casual users.
Key Takeaway:
WuDao, China’s version of GPT-3, is lagging behind among its competitors in terms of performance, as its responses are not expressive, natural-sounding, or effective. While the Chinese AI model was built on a significantly bigger Chinese dataset, it still has a long way to become an effective chatbot that can maintain coherent conversations.
