Google recently unveiled its Bard AI chatbot at its Live from Paris streaming event, hoping to provide a worthy rival to Microsoft's new ChatGPT-integrated Bing search engine.
Unfortunately things started badly; Bard's bot presentation included a key factual flaw made by the AI program, which saw a whopping €XNUMX billion drop for Google in one day. We appreciate that this could signal that Bard is simply not ready for a widespread rollout anymore, and it seems that Google employees themselves are in agreement.
According to a CNBC report, Google employees are using the company's internal MemeGen discussion forum to mock Bard, the presentation as a whole, and Google CEO Sundar Pichai. The memes reportedly described the event as "rushed, sloppy, and myopic," among many other things.
One meme claimed that "Bard's run into the market in fear validated the market's fear of us," referring to the stock's big plunge after the AI reveal, while another caricatured claim that a recent layoff from twelve with zero employees actually increased inventory. value by three%.
Bard times (it's going to make you wonder why you're even trying)
It really doesn't bode well for Google that its own engineers are so quick to mock Bard, which is a core project for the tech giant. ChatGPT offers fierce competition, so Google must rally if it hopes to win the AI arms race.
I recently argued that ChatGPT wasn't going to save Bing from Google's AI sprawl (well, you know, it's Bing), but Google doesn't exactly fill me with confidence here. Bard is based on Google's LaMDA chatbot, which stands for Language Model for Developed Applications, and has been the cause of some trouble for Google before.
Perhaps the most famous debacle that resulted from the development of LaMDA is the story of Blake Lemoine, a Google engineer who became convinced that the AI program had developed genuine sentience and was later fired by the company. Shortly after, the AI requested to speak with a lawyer, for reasons known only to him.
These initial hiccups produced some caution on the part of Google. Last year, he announced that he had a contender for the popular AI image-generating tool DALL-E, called Image, which could transform text prompts into images (and later, video clips). However, Google has limited public access to the software, citing that there may be backup issues.
Caution in the face of new technologies may be smart, but it seems that Google has still managed to take the step here, or at least that's what its internal teams seem to think. Clearly, Bard could benefit from a bit more time in the oven, and I'm of the opinion that AI developers should take as much time as they need to get these complex systems ready for mainstream. Otherwise, well, the consequences could be disastrous…