The AI chatbot race has begun in earnest – but one participant has already seemingly stumbled out of the gate: Google Bard. In the excitement and flurry of ChatGPT, Microsoft’s updated Bing, and others, Google announced that it will be participating with its own Artificial Intelligence offering, Bard, but an underwhelming launch event, followed by some mixed messaging, has left the search giant scrambling to catch up.
Microsoft debuted the newly ChatGPT-equipped Bing last month as an additional Bing Search feature. This is likely to try and capitalize on the impressive user sign-up numbers to OpenAI’s own ChatGPT tool, and drive users in the direction of Bing, as opposed to, say, Google.
Understandably, Google announced its own candidate – Google Bard. Now, leaked internal discussions at Google have offered some more insight into what Google Bard is supposed to be. According to an audio recording of a meeting with Google execs acquired by CNBC, it’s currently being developed as something other than specifically a search assistant tool. This is a notable difference from how it was first presented, including in Google’s own video of a demo – which did not inspire the greatest confidence in Google Bard after it returned incorrect information during the demo.
Not to be a ~well, actually~ jerk, and I’m sure Bard will be impressive, but for the record: JWST did not take “the very first image of a planet outside our solar system”. the first image was instead done by Chauvin et al. (2004) with the VLT/NACO using adaptive optics. https://t.co/bSBb5TOeUW pic.twitter.com/KnrZ1SSz7hFebruary 7, 2023
Confused messaging
This initial roll-out and subsequent internal messaging are somewhat at odds. In the above-mentioned meeting, executives answered questions from Google’s internal forum, most of which had to do with Bard. The product lead for Bard, Jack Krawczyk, answered one of these questions, which asked if search is the most appropriate purpose for tools such as Bard and ChatGPT.
This is a valid question because while they are large language text models (LLMs) that produce convincing and relatable human-sounding text, they do not ensure that their output is fact-based.
Jack Krawczyk responded with “Bard is not search.” He elaborated that Bard is an experimental product, and is more of a “sparkplug for imagination,” to help “explore your curiosity.” He did add that Google could not stop users from using Bard for search, but his emphasis is clear that this will not be its primary role and that if you want to search, that is what Google Search is for.
Elizabeth Reid, vice president of engineering for search at Google, backed Krawczyk up, echoing that Bard will be separate from search and that Google wants to “keep the heart of what search is.” She did add, however, that Google did have a prolific history of using large language models in search, perhaps indicating that Bard could become one of them.
Again, this is a slight departure from the lead-up to the announcement of Bard, which stated that Bard was being developed so that it could be integrated into Google search, and that there was a great amount of emphasis on this in the initial strategy. Several unnamed Google employees have said that this change in messaging has left them confused and that they found executives’ messaging inconsistent.
While Wall Street has not been too impressed with this mixed messaging (Google’s stock has dropped since Bard was announced), this is not necessarily all bad.
We are at the beginning stages of this technology, so in my opinion, caution is warranted and adherence to as much feedback as possible is sensible. Following Bard’s announcement, Google’s company leaders insisted that employees test Bard and revise wrong answers, quoting a “great responsibility to get it right.”
The position of ‘first out of the gate’ has already been taken, and interest in AI chatbots is bound to last for some time, so taking time to make sure it releases in the best possible state is a wise move Still, it’s hard to say when Bard will premiere exactly, as Google employees themselves are seemingly surprised by its current development.
Google’s present leadership has been under some scrutiny, especially for the recent Bard introduction, and nobody wants to show their long-term development cards. As Google has a monopoly in the search game, its AI chatbot is perhaps the most hotly anticipated, but as we have already seen, the slightest misstep can have big ramifications.
I have no doubt Google will try to avoid this happening again, but the sooner it gives more clarity to both potential users and its own employees, the faster it can catch up – and even overtake – its competition in the AI chatbot space. After all, Google is more used to leading the pack, than following it