The confusion has spread to Microsoft’s internal AI team. Reports say it is battling with funding restrictions and limited access to OpenAI technology. It has also muddled Microsoft’s controversial deployment of AI-powered Bing search last February, according to insiders.
Bing was shown to be vulnerable to prompt injection attacks at the time, disclosing business secrets and generating occasionally erroneous and insane replies to user queries.
According to the WSJ, OpenAI warned Microsoft “about the perils of rushing to integrate OpenAI’s technology without training it more.” They also “suggested Microsoft move slower on integrating its AI technology with Bing.”
A top concern for OpenAI was that Bing’s chatbot, Sydney, might give inaccurate or erratic responses. This early warning appears to have been easily ignored by Microsoft.
In a Wired interview published today, Microsoft CEO Satya Nadella stated that any problems with Sydney were merely part of Microsoft’s plan for teaching the chatbot to respond to real-world requests that couldn’t be tested in a lab.
“We did not launch Sydney with GPT-4 the first day I saw it, because we had to do a lot of work to build a safety harness,” Nadella told Wired.
“But we also knew we couldn’t do all the alignment in the lab. To align an AI model with the world, you have to align it in the world and not in some simulation.”
Taking Bing’s Thunder
As ChatGPT’s popularity expanded, some Microsoft workers expressed worry that it was taking Bing’s “thunder,” according to the WSJ.
Others wisely suggested that Microsoft may gain useful lessons from ChatGPT’s early public testing ahead of Bing’s launch.
Microsoft saw its partner rise in popularity by the minute. The corporation prioritized the rollout of AI-powered Bing to catch up.
The public took interest in the technology, and Microsoft wanted to be a greater part of the discussion.
The AI race victor was ChatGPT. It quickly drew the fastest-growing user base in history.
Introduced a month later, “The new Bing has yet to come close to the breakout success of ChatGPT,” according to the Wall Street Journal.
According to YipitData, ChatGPT “has nearly double the average number of daily search sessions as Bing search does.”
YipitData estimated that ChatGPT had 200 million monthly users, whereas Bing achieved 100 million daily active users in March.
ChatGPT and Bing received backlash for releasing chatbots that were capable of making up stories, lying, and other difficulties that many users were not expecting to experience. According to CNBC, OpenAI is still exploring on ways to prevent chatbot “hallucinations.”