Exploring Google Bard’s Botched Demo
by Adam Pease
Last week, Google debuted its conversational AI tool Bard. However, its public demo contained a notable error that many have pounced upon to critique the search giant. In this blog, we discuss Bard’s error and its implications for Google’s odds in the generative AI race.
Why Did Bard Fail?
In a promotional video released by Google, Bard can be seen being asked a simple question about the James Webb Space Telescope (JWST). In part of its answer, Bard makes an apparent mistake by suggesting that the JWST was the first telescope to photograph a planet outside the solar system. Of course, photos of planets outside the solar system have existed for years, but Bard’s ‘error’ is actually technically correct if one focuses on the phrasing a planet, in the sense that it did take the first pictures of a certain planet outside the solar system.
And yet, the Internet went into uproar over what appeared to be a major factual mistake. Even if technically correct, the answer was read by many to be flawed, and can be at best considered
a confusing and unclear response. The result was embarrassing for Google, with many pointing to its stock losses and suggesting that investors and shareholders were dissatisfied with the
Bard demo. Rumor has it that Google employees themselves have been gossiping about the botched launch, calling the release of Bard rushed.
What’s Next for Google AI?
Google is now locked in an intense battle with Microsoft to control the rapidly growing generative AI market. While it has many advantages, Google is coming late to the party, with
OpenAI’s ChatGPT leading many to sign up for Microsoft’s new AI-enhanced version of its search engine, Bing. Google, which has historically had issues bringing innovative products to
market, has leaned heavily on its search monopoly, which now appears to be in danger of a challenge from Microsoft’s aggressive plays in this emerging market.
Despite Bard’s bellyflop debut, Google still has a path forward for integrating AI into its services in a way that will give Microsoft a run for its money. ChatGPT makes many factual errors in its
outputs as well, so this release doesn’t indicate to us that OpenAI’s offering is objectively better. In fact, it provides an opportunity to acknowledge that these kinds of factual errors—or
‘hallucinations’ as they are sometimes called—are common in today’s large language model products.
Bottom Line
Ultimately, Google’s imperfect Bard demo doesn’t put it out of the running for generative AI, but it’s not the best start. This optics failure casts a shadow over Google’s place in the race; however, the race is still only just getting started.
See Adam LIVE on Tuesday, February 28, 2023 at 10 AM PT / 1 PM ET!
Leverage the Latest Generative Content Tools and Trends
Discussion of the generative content market has exploded in the past year as tools like ChatGPT and Stable Diffusion reveal the power artificial intelligence has to automate critical processes for content creation and business communication.
Emerging AI models make it possible to generate text, voice, images, code, and more, as many exciting developments in open source and emerging SaaS products have the market moving at a dizzying speed.
Join Aragon Analyst Adam Pease on Tuesday, February 28, 2023 where he will discuss how to leverage the latest trends and content in generative content.
This webinar with Adam will:
- Bring you up to date on the latest developments generative content
- Give you tools that your business can leverage right now
- Cover recent research about the trends that are shaping the direction of the market.
