![](https://cdn.mos.cms.futurecdn.net/G4nrodTaNvNmyuPQLTKihe.jpg)
Google Bard has been around for a few weeks now and we’re still trying to figure out how it separates itself from ChatGPT and Microsoft’s Bing with ChatGPT. It looks like Google is too, as we’ve seen rumors that Bard could start invading your text messages, be integrated into Search and maybe even screen your calls one day.
Now though, we should start to get a clearer picture as to how Google is changing Bard as it goes along. Google has unveiled “Experiment updates” a page in the Google Bard user interface that highlights what recent changes have come to Google’s chatbot. At the moment, the changes are sparse, but there are a few headlines from Bard’s first update.
Experiment update 2023.04.10: What’s coming to Bard
Some of the updates we already knew were coming. In the subheading “Updates to Bard’s capabilities,” Google says that they’ve made Bard smarter, with upgraded math and logic processing capabilities.
This upgrade was tipped a week ago and comes from Google integrating its more powerful PaLM (Pathways Language Model) model with the LaMDA (Language Model for Dialogue Applications) model that currently powers Bard. Google says the reason behind this change is that Bard seems to be struggling with math and logic prompts, and a more logical AI model will certainly be required if Bard wants to compete with OpenAI’s GPT-4.
The other big change is that Google is upgrading Bard’s “Google it” feature. When you give Bard a prompt, it returns with an answer, but it will also give you the option to Google it if you want to see a possible Google Search query related to the topic.
![](https://cdn.mos.cms.futurecdn.net/EDfHKdfPHJtQ6AVPndHiy4.jpeg)
Now, clicking this Google it button will provide several search terms that you can click to continue down your rabbit hole on Google Search rather than directly in Bard. In theory, this should direct you more accurately to your desired information, but it could also have the added benefit of reducing the likelihood of “AI hallucinations” (when chatbots go off-script) as users could terminate Bard chats quicker with more relevant Search options available to peruse.
Frankly, the most interesting thing about these Experiment updates is the use of the word “Experiment.” Google has been keen to highlight that Bard gets things wrong and is far from a polished product. This is in no small part due to the $100 billion mistake Bard made at its debut and Bing with ChatGPT going off the deep end and spooking potential AI users. Doubling down on this experiment label could allow Google to shrug off future Bard snafus.