With the chatbot even plagiarizing our content in the Google Bard vs ChatGPT comparison, things were looking glum. However, at the Google I/O 2023 devcon, the company has made short work of its opponents with the now-revamped Google Bard AI with many new features. But what are all these amazing new improvements? Let’s check them out below.
Previously Google Bard was powered by a small portion of LaMDA, another one of the company’s large language models (LLM). However, that has now completely changed as Google Bard is now powered by PaLM 2, the company’s next-generation language model. This relatively new LLM is now improved and comes with upgraded reasoning and coding changes.
The inclusion of PaLM 2 provides Google Bard with new features that were not possible before. In its blog post, Google mentions that the new LLMs wide-ranging dataset includes scientific papers and mathematical expressions. This makes it even more adept at complex logical reasoning and mathematical problems. PaLM 2 is also faster than previous models, so expect quicker replies from Bard from now on. The new LLM packs a lot of other advancements.
2. Google Bard Can Now Read Images
While Bard isn’t multimodal like OpenAI’s newest GPT-4 LLM, it will now be able to intelligently read images and respond back. This is going to be possible by direct integration of Google Lens into Bard. For those unaware, Google Lens is an image recognition technology by the company. This integration should make it possible for Bard to read all types of images and draft responses.
Google showcased the same when it fed a picture of two dogs to Bard. The AI was then asked to “write a funny caption about these two.” Within just a few seconds, Google Bard analyzed the image and wrote a few creative captions ready for use. While the company did not show other use cases, Google Bard image recognition is definitely one of the best features coming to the bot.
Companies like Microsoft have had tremendous luck by bringing native services like the Bing AI Image Creator to their portfolio. Combined with the best AI art generators, these tools are almost impossible to live without. While Bard has been behind in this particular aspect, that is going to change soon. Starting soon, Google is bringing a built-in image generator into the bot, so users can seamlessly create AI art without leaving the tab.
This is made possible by Adobe Firefly, “a family of creative generative AI models” that can generate high-quality images on the fly (pun intended). Google is soon going to integrate Firefly’s model right into Bard for easy access to the service. This means users will be able to request Bard for images through prompts and will get image outputs through Adobe Firefly.
You will even be able to edit these images further or pull them into Adobe Express for more editing. With AI image generation tools like Midjourney quickly becoming the norm, a built-in image generator is a welcome feature in Google Bard.
When Microsoft Bing AI Chat went open to all, it announced quite a lot of new features including the ability to respond back with images. Google has followed its rival’s suit and empowered Bard with visual responses. Starting soon, when asked across different queries, Bard is going to intelligently use Google Search and provide helpful visual context along with text results.
It will display these combined results right in the chat itself for easy access. According to Google, these images should give users a better sense of their results and we agree. While not live right now, image insertion in Bard should be available to users in the coming weeks. It will also come to Google Search with the new “Generative AI” feature.
Besides Google Bard being a chatbot, the company is slowly planning to enhance the bot even further by integrating other services into its chatbot. In an effort to streamline workflow, Google is creating a seamless connection between Bard and other services, such as Google Sheets, Docs, and even Maps. This will allow all these services to work in tandem and breathe even more new features into Google Bard.
The company displayed one of these examples in its keynote recently. Sissie Hsiao, the General Manager for Bard showcased the same when she asked the bot to find different college programs based on her interests. Once Bard was done with the results, Sissie requested Bard to show them on a map and the AI displayed them instantly using Google Maps.
However, she then asked it to create the results in a neat table full of different columns for college names, locations, degrees, and more. Once done with the table, Bard was able to export the entire table to Google Sheets while keeping the formatting intact. This is a big step towards Bard helping people collaborate across platforms. While not all of them are live just yet, Google is planning to bring the said features to Bard soon.
Well, the service integration does not end with Google products, as extensions from other companies will also help make Bard AI way more useful. If you have been following AI developments recently, you might remember that OpenAI’s ChatGPT now supports plugins. This was hailed as a big step in AI as the bot could now be powered with even more experiences. Well, Google hasn’t been left behind. Besides Adobe Firefly’s inclusion, Bard is getting its own plugin experiences through partner companies.
Starting soon, Google will be partnering up with services like Kayak, Spotify, OpenTable, ZipRecruiter, Instacart, Wolfram, and Khan Academy to bring selective experiences to Bard. This will help include even more Google Bard features and will open possibilities for education, productivity, leisure, and much more. Combined with Google’s integration with its own services, the upcoming Bard tools will empower the AI bot to do even more.
7. Bard AI Can Code for You
Previously, when we compared Google Bard against ChatGPT, we found that the former could not furnish any coding requests. Even Bard AI’s FAQ page at the time confirmed that the company hadn’t added support for that yet. While the bot could still render out basic HTML responses, the code was at times incorrect and could not be relied on. However, Google I/O 2023 has now revealed that Google Bard AI is getting programming features.
Google Bard can perform a plethora of coding features, including the ability to generate, debug and explain code without problems. Demoing the same at I/O, Bard AI seamlessly generated the “Scholar’s Mate move in Chess” using Python and even provided an in-depth explanation for the code. Google Bard will even provide citations to rendered code for double-checking.
Once done with the output, users will be able to export the code to either Google Collab or Replit for producing results. Google Bard will support 20+ programming languages, including HTML/CSS, Java, C, C++, Dart, Kotlin, and more. This will be a big help to people new to coding and previous users who were learning how to write code with ChatGPT.
While ChatGPT can dish out excellent responses quickly, it provides a limited set of answers, which can feel repetitive at times. However, Google Bard has and still possesses one of the best features in the form of drafts. Put simply, drafts in Google Bard give users the option to get multiple responses from the same query given to the AI chatbot. Each output response is different from the other and as such might have a different take depending on the prompt.
By default, Google Bard writes three different versions of the same answer. This gives people the freedom to choose from the response which suits their query best. You can either pick one or mix & match them to get the answer you were hoping for. Furthermore, there are no special steps required for drafts as the AI bot automatically creates these answers for you. So, for those looking for different takes given by the AI, the draft feature is a welcome one in Google Bard.
Even though OpenAI’s GPT-4 LLM comes packed with a ton of features, including multimodal response, it’s still limited to just more than 26 languages. While that is quite a lot, it doesn’t even begin to cover the expanse of languages offered worldwide. Google
This new feature dramatically increases the scope of the AI bot as it should soon be able to fully converse with users over the globe in the language of their choice. These 40 languages will include all the widely spoken ones for greater support. Moreover, Google mentioned that users will now be able to talk to Bard in Japanese and Korean. Once fully rolled out, extended multilingual support will be one of Bard’s best features.
If you’ve followed our guide on how to save and share ChatGPT conversations, you already know how annoying it becomes to copy and paste text from AI chatbots to other apps and services. And while Google is now providing an export button for its coding outputs, the company is taking a further step when it comes to sharing and using outputs.
Starting now, you will be able to export your Google Bard conversations directly onto services like Google Docs and Gmail. This applies to all kinds of messages by Bard. Google mentions these one-click options should simplify workflows and make the entire AI output process even faster.
The new export chat feature is seamless as when Bard is asked to export an e-mail format onto Gmail, it quickly opens a new window and pastes the output for immediate submission. This is the same case with Google Docs. Users looking for a way to quickly export their chats will find solace in this feature.
While some users might find this feature a bit too tiny to include, believe us when we say that the crowd at Google I/O thought otherwise. Starting now, Google Bard is available in, well, an all-new dark mode. As it goes with dark modes, it’s a good mix and match of different shades of dark and grey and looks appealing.
To switch over to the new dark mode, all one needs to do is click on the handy dark theme button on the bottom left sidebar and they are good to go. So go on and enable one of Google Bard’s most welcomed and best features.
