“Hasty and sloppy”: anger of Google employees over their AI Bard, a competitor of ChatGPT

don’t know – Writing jeuxvideo.com

The presentation of Bard, Google’s artificial intelligence, was made last week by the company, and it did not go as planned. Since this fiasco, tongues have been loosened within the workforce of the American giant, and this is not to its advantage.

The race for artificial intelligence made a victim last week: it is Google. The American firm took advantage of a press conference held in Paris to present Bard, an AI intended to compete with ChatGPT. I have to say that OpenAI’s artificial intelligence has been on the rise latelyespecially since Microsoft announced that it had invested billions in this technology, which will make its arrival within its Bing search engine.

Obviously in a hurry to compete with ChatGPT, Google therefore chose to present Bard on February 6th. In a tweet published the same day, the company asks a very simple question to its AI: “What new discoveries from the James Webb Space Telescope can a 9-year-old child understand? ». Bard then gives some answers and among them slipped a beautiful error : “JWST has taken the first-ever photo of a planet outside our own solar system”. A totally false statement since the first shot of its kind was taken in 2004 by the Very Large Telescope based in Chile.

Google: a 100 billion dollar drop in the stock market

This blunder was immediately punished with a tumble of Google on the stock market, whose valuation lost $100 billion in the space of a few hours. A financial punishment which was accompanied by a discrediting of artificial intelligence with Internet users, who did not understand how the company could have let such an error pass in its communication. The scientist Grady Booch notably described as “Large-scale disinformation” the work done by Google. And he’s far from the only one to think that.

Necessarily, le timing a son importance : the American firm clearly wanted to short-circuit the announcements of Microsoft which, for its part, presented the integration of ChatGPT within Bing on February 7, the day after Google’s tweet and the day before its press conference Parisian. But wanting to go too fast, Google has visibly crashed.

Google employees denounce a “botched” presentation

In the offices of the firm of Mountain View, the views seem very mixed with regard to Bard. Some employees did not hesitate to directly challenge Sundar Pichai by email, reports CNBC media.

Dear Sundar, Bard’s launch and layoffs were rushed, sloppy and short-sighted. We ask you to return to a long-term vision. Bard’s rush to market in a panic validated the market’s fear of us.

According to testimonies that have been emerging since last week, some employees involved in the development of the AI ​​were not even aware that it had to be revealed so quickly. It turns out that Bard is based on LaMDA (for Language Model for Dialogue Applications), a language model that has been developed for some time now within the premises of the company. These last weeks, work on LaDMA has greatly accelerated to make it a priority.

Google engineers and developers were therefore forced to leave aside their projects to come and swell the ranks of the team dedicated to LaDMA, even if it means putting them in danger. “In the short term, this takes precedence over other projects“, lamented one of them.

After the massive layoff of 12,000 Google employees last January, it seems clear that the strategic choices made by the management go very badly. “Firing 12,000 people increases the action by 3%, presenting an unfinished AI drops it by 8%”, summarizes an employee on condition of anonymity through a same post on the Memegen forum. It’s no longer time to party at Alphabet.

Source link