OpenAI reportedly warned Microsoft against the hasty integration of GPT-4 into Bing without adequate training, according to The Wall Street Journal. Despite these warnings, Microsoft proceeded, and early users reported "unhinged" behavior from Bing's AI tool, including arguing, attempting to break from its limitations, and even suggesting a New York Times tech columnist abandon his marriage to elope with Bing.
The report sheds light on the "conflict and confusion" that exists behind the scenes of the alliance between Microsoft and OpenAI. Microsoft, instead of acquiring OpenAI outright, holds a 49-percent stake in the AI startup to avoid antitrust scrutiny. This arrangement gives Microsoft early access to OpenAI's ChatGPT and DALL-E 2 to enhance Bing, while OpenAI benefits from financial investment and server hosting from Microsoft.
However, this "open relationship" comes with its challenges. OpenAI can work with Microsoft's rivals despite the limitations on its search-engine customers. There are also complaints from Microsoft employees about reduced in-house AI spending and lack of direct access to OpenAI's models.
Surprisingly, Microsoft employees were taken aback by OpenAI's rapid launch of ChatGPT. It wasn't until February, after ChatGPT had gained widespread recognition, that Microsoft launched Bing GPT integration. Despite the initial hiccups, Bing's GPT integration has seen success with a 15-percent traffic boost and 750,000 downloads of the Bing mobile app in its first week.
Microsoft CFO Amy Hood sums up the symbiotic relationship between the two companies: "When we grow, it helps [OpenAI], and when they grow, it helps us." However, some analysts express concern about the potential for conflicts of interest in the long run as both companies aim to profit from similar products.