Can Google save its golden goose or will it just kill it trying? That’s the question behind the launch of the company’s chatbot Bard, hastily announced after the overnight success of ChatGPT in early 2023.
With Bard, Google has to walk a tightrope: offer users an experience that can compete with AI-powered Bing Chat and ChatGPT without cannibalizing its hugely profitable search business in the process.
And it has to do all of that under the kind of scrutiny that an upstart competitor can avoid, but a market leader has to tackle head-on. It’s an interesting quirk when ChatGPT “freaks out” on false information, but it’s a much different feeling when AI backed by the world’s third-largest company does the same.
At the bottom of every Bard conversation is a disclaimer: “Bard may display inaccurate or offensive information that does not represent the views of Google.”
For a company that once proudly described its vision as “organizing the world’s information and making it universally accessible and useful,” it’s an odd reduction in ambition, like Tesco launching a new range of products that can cause food poisoning or Ford warning that its last the car may not stop when the brakes are applied.
That juxtaposition is why it took Google so long to bring Bard to market in the first place. The company has long been leading the way in AI research, with ChatGPT based largely on research that emerged from Google Labs (the “T” in GPT stands for “transformer,” a technology invented by Google). But his inability to create versions of a “big language model” that wouldn’t “blunder” with facts or fall victim to “alignment” issues raised fears of backlash if he released the systems publicly, until ChatGPT forced his hand.
Google says that it has built Bard with a focus on alignment, ensuring that the chatbot behaves in a safe manner. But the nature of the technology makes it difficult to guarantee that, and unlike a bad search result, the company will fight to blame anyone other than itself for the damage caused by those errors.
So why release it? Because the excitement around ChatGPT and Bing Chat has reached a fever pitch, and Google can see the writing on the wall: If users start switching to chatbots for a few queries, they may never return to search engines. It’s better to offer a chatbot itself and tempt users back to conventional search when it’s better, which is why every Bard conversation has a prominent button to request a Google search.
With an extra month of preparation and evidence of Microsoft’s disastrous initial attempt to control Bing Chat, it seems unlikely that Google will have any immediate disasters. But be careful what you wish for, as a world where users switch millions from Google Search to the Google Bard, which costs far more per query and makes no money to operate, is hardly better for the bottom line than losing them. . to Bing in its entirety.