The launch of ChatGPT shook up several tech companies. Google, whose revenue depends heavily on its search activity, something ChatGPT could eventually threaten, was particularly concerned.
Now, less than three months into ChatGPT’s existence, Google has announced the launch of a ChatGPT-style AI called Bard to take over ChatGPT’s seemingly unchallenged reign. But how will Bard work? Will Bard be better than ChatGPT? Here’s everything we know so far.
- Advertisement -
What is Google Bard AI?
Bard is a Google-owned experimental AI language model that can generate human-sounding responses to prompts or input provided. It is based on Google’s powerful conversational language model, LaMDA (Language Model for Dialogue Applications).
Bard was announced by Google CEO Sundar Pichai on Google’s The Keyword blog, finally putting a face to a long-rumored product Google had in the pipeline to rival ChatGPT.
Unlike ChatGPT, Sundar Pichai explained that Bard will initially only be available to a few trusted testers, with plans to scale to more users. The announcement also showed what the initial version would look like with several examples, some of which look strikingly similar to what ChatGPT would do. So, will Bard be just another ChatGPT?
Why Google Bard might be better than ChatGPT
ChatGPT was released on November 30, 2022, after a long period of development. The entire development process, including the underlying technology that powers it, took several years. At first blush, that would mean Google had less than three months to create a product that rivals ChatGPT, a product built on years of evolving technology.
However, the reality is different. Google has been investing heavily in artificial intelligence fields such as natural language processing for several years. In an apparent twist of fate, while Google is trying to rival ChatGPT, the very technology architecture that powers ChatGPT, known as the Transformer architecture, is the brainchild of Google research. In other words, Google is now trying to overcome a business threat technically based on Google technology.
In an apparent reference to ChatGPT, Sundar Pichai was quick to point out its pioneering status in building Transformer models in his Bard announcement blog post. He pointed out that Google search is “the foundation of many of the generative AI applications” currently available today.
However, beyond his words, the evidence is clear. This is why LaMDA, the technology that powers Bard AI, is very similar to GPT-3, which powers ChatGPT. So, no, Google isn’t a newcomer to the race to create a general-purpose AI chatbot. In contrast, the company has been building similar, if not better, technology for years and may enjoy a significant advantage over OpenAI on who ultimately becomes the parent.
But that is not all. Google also enjoys the benefit of having more data to play with. In the AI business, especially when it comes to training conversational models like GPT-3 and LaMDA, more data could mean better results. It is not entirely clear if and how Google intends to integrate real-time data or information from the web into Bard’s responses. However, if that happens, it would be a significant advantage over ChatGPT’s pre-trained approach to responding to requests.
In simpler terms, this would mean that Google’s Bard would be able to provide fresh and up-to-date answers while ChatGPT, unfortunately, would be limited to event-related information, no later than 2021 (ChatGPT’s current data training shutdown).
How will Google Bard work?
From the demonstrations provided by Google, it’s not entirely clear whether Bard will exist as a stand-alone fast-response web interface like ChatGPT “after the initial testing phase.”
However, in typical Google fashion, Google will almost certainly push to deeply integrate Bard into most of its products, including Google Search, similar to what Microsoft is trying to do with Bing and other Microsoft products using ChatGPT technology.
Google’s Bard demonstration presents factual issues
But despite the interest and understandable hype over Google’s Bard announcement, its first demonstration didn’t go well. Eagle-eyed astronomers were quick to point out that at least one of the “facts” provided by Bard was dead wrong. For example, Bard said the James Webb Space Telescope “took the same pictures of a planet outside our solar system,” but the European Very Large Telescope got that accolade in 2004.
Bard is still in its very early stages, but its false claim was a clear indicator of problems integrating conversational AI technology deeper into a general internet search, where many will trust what is written.
Will Google’s Bard AI be better than ChatGPT?
Google has sat on a huge array of truly phenomenal AI tools, only periodically giving the public a taste of what it has to offer.
With OpenAI’s bold gamble to open ChatGPT, with all the associated risks, to the public, Google could finally shed its fear of reputational damage and show the world what it has to offer. If the glimpses of Google’s AI models we’ve seen over the years are anything to go by, then ChatGPT might indeed have a worthy competitor in the Google Bard.