Google Enhances Bard with ChatGPT Insights, According to AI Document Leaks

In 2023, Google aimed to enhance its chatbot Bard, later rebranded as Gemini, amid challenges from ChatGPT. Internal documents revealed that Google’s contractors at Scale AI leveraged ChatGPT’s outputs to refine Bard’s responses, which were initially criticized as “rushed” and “botched.” Scale AI generated thousands of responses from ChatGPT, comparing them with Bard’s to improve the latter’s performance. Contractors were incentivized to craft answers that could surpass those of ChatGPT. Scale AI maintained that these activities merely comprised standard performance evaluations, not training, and Google affirmed they did not use ChatGPT’s data for that purpose. Despite this, experts noted that evaluation can still influence AI models. Additionally, a publicly accessible document detailed sensitive information about the contractors and their work, prompting an investigation by Scale AI. Following these efforts, Google successfully reemerged in the AI scene, launching over 100 new products and features.

Source link