Bard Unbarred

Much ink has been spilled in recent months over Google’s apparent hesitancy to release an artificial intelligence chatbot to the general public. Once seen as a leader in the field, many view this as a major miscalculation by the company, and are now even questioning Google’s future dominance in their core business areas, like internet search.

With OpenAI’s ChatGPT making huge waves in the business world and garnering more users by the day, Google decided to make their move in early February of this year. In a livestreamed event, they unveiled their own large language model-based chatbot called Bard. Unfortunately for Google, this event did not go exactly according to plan. When Bard was asked about discoveries made by the James Webb Space Telescope, viewers noticed that the answer given, which mentioned that the telescope captured the first pictures of a planet outside of our solar system, was incorrect. Ouch!

This poor performance led many to believe that Google had in fact lost its luster, and the stock price quickly sank by eleven percent, reflecting that loss of confidence. Now Google is hoping to get a do over, and they have announced just today that Bard is officially being rolled out to users, albeit in a very limited way at this time. Those in the U.S. and the U.K. can join a waitlist to be among the first to try Bard. There are no further details available about what the timeline looks like, or how many users are being accepted initially, so exactly what it means to be on the waitlist is unclear.

In all fairness to Google, Bard’s misstep was hardly the first time a chatbot had made a factual error. Anyone that has spent some time interacting with the best models presently available to the public will have noted that these errors are by no means unheard of, so it would be unfair to write Bard off for one mistake, even if it was highly visible. And more available options tends to be better for consumers, and also to drive innovation, so we should all be pulling for new entrants into this space.

Bard is being released as an experiment, to see how it helps people to boost their productivity, and how it can help to fuel creativity. Google is also looking for user feedback to help them improve the lightweight and optimized version of LaMDA that powers this chatbot. With the feedback given, they intend to update Bard’s underlying model with newer, more capable versions in the future.

The features built into Bard give some hints as to how Google may incorporate it into future products. Responses generated by the chatbot have “Google it” links embedded within them to make it simple to run a traditional Google search to verify the truthfulness of statements made by Bard. The Google team describes Bard as a complementary experience to Google Search, but also note that they are planning to integrate large language models into their search tools in a deeper way in the future. Perhaps before too long a chatbot will be our starting point, with traditional search engines complementing that experience. Only time will tell — but at the pace the field has been moving, that time may be very soon.

To get yourself on the waitlist, you can sign up here.