Recently, Google's competition in the generative artificial intelligence field has drawn attention again. According to documents exposed by Business Insider, Google contractors are accused of using ChatGPT to train and improve the performance of its AI assistant, Bard (now called Gemini). This news suggests that Google may be leveraging a competitor's technology to enhance its own product, which is noteworthy.

Sundar Pichai, CEO of Google's parent company Alphabet, once expressed dissatisfaction with Microsoft's collaboration, emphasizing that Google has not lost opportunities in the AI field. However, newly disclosed documents show that contractors compared Bard's outputs with thousands of ChatGPT responses to improve Bard's answer quality, aiming to match ChatGPT. Satya Nadella, CEO of Microsoft, mentioned that OpenAI had a two-year lead advantage in developing ChatGPT, posing a challenge for competitors like Google.

Leadership at Scale AI quickly responded to these matters, emphasizing that they have never used ChatGPT responses to train Gemini or other models. They stated that the relevant documents were merely "standard comparative evaluations," a common practice in the industry, often misunderstood as actual model training activities.

At the same time, Google's relationship with Scale AI faces new tests. Meta plans to acquire 49% of Scale AI's equity for $14.3 billion, valuing Scale AI at $29 billion. Notably, Google was originally Scale AI's largest client, planning to pay up to $200 million by 2025 to obtain its crucial manually labeled data for Gemini development. However, Meta's acquisition plan seems to have made Google unhappy, prompting it to consider terminating its partnership with Scale AI.

This series of events not only reveals the competitive pressure faced by Google in the AI field but also draws attention to the complex relationships of cooperation and competition among industry giants.