Report Details Internal Conflicts Over Microsoft’s AI Push

A Wall Street Journal report details the conflict and confusion that resulted when Microsoft decided to go public with its AI-powered Bing chatbot back in February.

There’s a lot to this report, but the most interesting bit, to me, is that OpenAI, which makes the ChatGPT AI technologies that power the Bing chatbot, warned Microsoft that it needed to take more time to address the inaccurate and bizarre responses its chatbot generated. Microsoft ignored those warnings and the resulting backlash threatened to undermine the quick burst of positive media spin that it had initially received.

Many—myself included—suspect that Microsoft had rushed Bing chat to market in a bid to improve the usage of its ad-supported search engine: after all, OpenAI’s ChatGPT had quickly surged to 100 million active daily users, and that kind of bump could help make Bing financially more viably. And while Microsoft did report that Bing hit that same 100 million daily active users mark a month after it announced the Bing chatbot, the Wall Street Journal says that Bing usage still lags ChatGPT by a long shot.

“The new Bing has yet to come close to the breakout success of ChatGPT,” the publication notes. “ChatGPT has already reached 200 million monthly users, according to analytics firm YipitData, making it one of the fastest-growing consumer apps in history. On computers, it also has nearly double the average number of daily search sessions as Bing search does.”

Inside Microsoft, there was dissent about the Bing chatbot and ChatGPT. Some worried that the rapid success of ChatGPT 4, released last November, would overshadow Bing (and given the numbers above, that’s exactly what happened). They argued that the software giant should hold off and learn from how customers were using ChatGPT before coming to market. But others complained that Microsoft’s non-exclusive access to OpenAI’s technologies would come back to haunt it as competitors emerged after Microsoft had slowed investments in its own in-house AI.

And that has happened, too. The problem with OpenAI is that Microsoft has what the Wall Street Journal calls “influence without control”: It has invested $11 billion in the firm, but OpenAI is free to partner with other companies and to sell its products directly to customers, and it has done both. Indeed, the Microsoft and OpenAI sales teams sometimes pitch the same customers, the Wall Street Journal says. Typically, Microsoft would have acquired OpenAI to keep the technology for itself. But with OpenAI, Microsoft deliberately tried to avoid antitrust scrutiny by providing only 49 percent of its investment funding.

Microsoft has also turned on partners that wish to use AI in search products that might compete with Bing: In March, it alerted them that they could not use Bing search results in AI-based products that compete with Bing AI. And with OpenAI reaching out to various search engine providers over the past year, Microsoft has not coincidentally raised prices on those that license Bing. DuckDuckGo, for example, partnered with OpenAI and used ChatGPT and Bing to power its DuckAssist chatbot. But it was forced to withdraw that product from the market several weeks after the launch because of Microsoft’s new fee structure.

Where the Microsoft/OpenAI relationship ends up is clear. For now, of course, executives from both companies speak warmly about each other, and it’s fair to say that OpenAI needed Microsoft’s investment as much as Microsoft needed OpenAI’s technologies. But there may come a time when this relationship becomes a bit more one-sided. And it’s hard not to imagine that it will be Microsoft that’s left wondering what went wrong.