It has been a few months since Microsoft announced its Bing AI, and slightly over a month since the company removed the waitlist associated with it. You’ve probably seen plenty of examples – maybe even experienced some yourself – of the generative AI tech giving very weird responses to prompts. Few would fault you for thinking that the tech still needs polish before being rolled out. It appears that the company was aware but done it anyway, according to a new report.
To revisit some background detail, OpenAI launched the free trial of ChatGPT late last year, while Microsoft only announced Bing AI earlier this year. The Wall Street Journal reports that the former warned the latter “about the perils of rushing to integrate” the latter’s tech without more training. This is because while the free version of ChatGPT is based on the existing GPT3 language model, Bing would use the then-unreleased GPT4.
OpenAI warned that it still needed “time to minimise issues like inaccurate or bizarre responses”. Instead, the report noted that some of Microsoft’s people “were worried that ChatGPT would steal the new Bing’s thunder”, thanks to the timing of the release. And so the few people who got early access to it reported as they did, emulations of wild emotional swings and all.
Overall, it paints a picture of a strange working relationship between the two companies. As the situation between ChatGPT and Bing AI illustrates, while Microsoft and OpenAI have indeed partnered with each other for AI endeavours, both companies still had their own AI projects going on. And while both companies are using each other’s tech, the report gives the impression that they are still trying to one up each other on that front. Ultimately, the report cites analytics firm YipitData as saying that, at least on PC, ChatGPT “has nearly double the average number of daily search sessions as Bing does”.
(Source: WSJ)
The post OpenAI Warned Microsoft About The Possibility Of Bizarre AI Responses appeared first on Lowyat.NET.
0 Commentaires