Google Bard chatbot will be available in 180 countries, will get new language model PaLM 2, dark theme and visual search

During the Google I/O conference, the company’s CEO Sundar Pichai announced a significant update to Bard chatbot, which works on the basis of a large language model. New features include support for Japanese and Korean languages, simplified text export to Google Docs and Gmail, visual search, and a dark theme.

Importantly, Google is also canceling the waiting list for Bard, making it available in English in 180 countries, but there is still no access to the chatbot when entering from Ukraine. Maybe Google hasn’t had time to open it to all users yet, but it’s worth noting that the company hasn’t published a list of countries where Bard will be available.

Among the future features of the chatbot, which Google is only planning to implement, are the generation of artificial intelligence images with the help of Adobe and integration with third-party web services such as Instacart and OpenTable.

Google emphasizes that Bard is still an experimental project and is not intended to replace its search engine. However, Bard has not yet generated the same level of interest as its competitor, OpenAI’s ChatGPT and its derivative, Microsoft’s Bing chatbot. Google aims to address this issue by integrating numerous new features and modernizing Bard to use the new PaLM 2 language model, which is expected to improve its performance and overall usability.

Google claims that the improved Bard does a great job writing code, including debugging and explaining code in more than 20 programming languages. Therefore, some of the improvements focus on this functionality, such as a new dark mode, improved code quotes (which will provide sources and explain snippets), and a new export button. The latter can port code to Google’s Colab platform, and will now also be compatible with another browser-based IDE, Replit (originally for Python queries).

For broader use, Google is adding more visual capabilities to Bard, allowing it to analyze images, include images in search results, and generate visual effects using artificial intelligence. The latest feature, which is expected to launch in the coming months, will be powered by Adobe’s Firefly software.

The visual results in Bard will be similar to the results of some Google queries. For example, if a user asks: “What are the must-see attractions in New Orleans?”. Bard will generate a list of points of interest, such as the French Quarter and the Audubon Zoo, accompanied by relevant images.

An interesting feature is the ability of the system to generate responses based on the image. This feature, powered by Google Lens, identifies objects in photos. For example, if a user uploads a photo of their dogs and asks them to add a funny caption to it, Google Lens detects the dog’s breed, allowing Bard to generate a caption related to their characteristics.

Google also plans to integrate Adobe’s Firefly image generator with artificial intelligence into Bard. Adobe promotes it with a focus on the ethical use of its training data, which is controversial and leads to lawsuits against other AI tools for creating images such as Stable Diffusion. It’s the first of many third-party integrations for Bard, with the company promising direct connectivity to apps “from Google and other amazing services.”

While these improvements are significant, Google is essentially duplicating features already offered by its competitors. In March, Microsoft integrated OpenAI’s DALL-E image generation capability into Bing. In addition, the latter also announced plug-ins for ChatGPT at the beginning of the year, in particular, an extension that allows the chatbot to receive up-to-date information from the Internet.

Despite these improvements, Bard’s ultimate purpose remains unclear. Although Google claims that the chatbot is not intended to replace the search engine, this does not prevent users from using it as a replacement for Internet searches. Perhaps Bard will become a testing ground for Google’s new developments in the field of generative artificial intelligence. However, it can also simply be a parallel service, because Google often does not focus on any one product. Just remember how many messengers the company had, and it still offers several applications for video calls at once.