Ever since its launch, Google Bard has gone through numerous iterations and updates based on user feedback and developer suggestions. This includes the chatbot’s transition to a more capable and robust PaLM 2 language model, which the company has also deployed in its experiments toward building a capable medical chatbot. Bard has since progressed to become a refined platform that has come to compete with its arch-rival ChatGPT on a more level footing, with numerous features being added to the chatbot at regular intervals. Bard continues to add users at a steady rate and remains a potent competitor to OpenAI’s offerings. With Bard, Google seeks to maintain its position as the other pole in the extant duopoly that persists in the AI chatbot market. Apart from the updates, Bard has also undergone considerable improvements in its responses, making it easier for users to interact with the chatbot and explore Bard’s capabilities.
Given that Bard did not have a promising launch in its initial stages, its improvements have worked to bring the chatbot back into the limelight, with more favorable opinions pouring in. From fixing accessibility issues to enhancing the underlying model, Google has worked carefully to raise the bar with its AI chatbot. As competitors like Bing continue to collaborate with ChatGPT to service their AI search engine needs, Google, too, is aiming toward understanding AI behavior and integrating key aspects of it with the firm’s extant search platform. The upcoming sections explore the latest upgrades added to Bard, shedding light on its various enhancements vis-à-vis previous versions.
Google Bard’s Key Update: An Overview of the New PaLM 2 Model
The new language model has extended Bard’s capabilities, making it far more capable than previous versions.
Google Bard’s major transition came when the chatbot developers moved the framework from a lightweight derivative of the LaMDA language model to the more advanced and robust PaLM 2. The newer LLM is vast and its parameter size touches a gargantuan 540 billion, making it one of the most extensive data profiles used by a language model. PaLM 2 is also multilingual, with the training data sourced from over 100 languages, making it a potent competitor even to GPT-4—OpenAI’s most advanced flagship model. Google is also developing other customer-facing AI products and solutions using the PaLM 2 model alongside Bard. This includes Search Generative Experience, which is an experimental generative AI search engine being tested and developed by Google. Apart from its impressive linguistic capabilities, PaLM 2 is also stated to possess advanced reasoning as well as mathematical and coding skills. All of these attributes add considerable weight to the current version of Bard, making it a heavy hitter in the persistently competitive AI chatbot realm.
PaLM 2 has passed several language proficiency exams and has gained “mastery” level expertise in its results. Its dataset has also spanned numerous scientific research papers and mathematical problems that have allowed the model to train better on common sense reasoning. Moreover, these capabilities allow Bard to provide more accurate answers in response to complex and detail-oriented user queries. The new Bard is also adept in Python and Javascript after being trained using open-source code databases. This has allowed Google to launch Codey—a dedicated coding assistant based on generative AI capabilities, designed to aid developers in their efforts. With this model forming the foundations of the new Bard, it is evident that the chatbot also provides enhanced coding support to its users. With its multilingual profile, users can also utilize Bard for a variety of translation tasks with relative accuracy.
Additional Features and Google Bard API
Bard’s features are being optimized across different platforms to enhance interoperability.
Apart from the major upgrade to a better language model, several additional features and capabilities have been added to Google Bard following its launch. With enhanced linguistic proficiencies provided by PaLM 2, Google Bard is now accessible in over 40 different languages and a larger number of locations across the world. Apart from the language developments, Google Bard has also become multimodal, with Google integrating its image identification service—Lens—with the chatbot. This also comes as a key development in the constant rivalry between Bard and ChatGPT, with GPT-4 also having similar attributes. Currently, Lens supports only the English language on Bard; however, its features should be extended to other languages in due time. Google has also begun providing text-to-speech support, with the chatbot being capable of reading aloud responses in 40 languages. This enhances Google Bard’s credentials when it comes to assistive technologies by improving accessibility features.
Apart from these useful updates, Google’s LLM venture has also embarked on an API project to compete with OpenAI’s API services. The feature, which is still under beta testing, is undergoing focused trials to refine and gather user opinion. Essentially, Google API will allow users to connect the rich features of Google’s language models to external applications and third-party platforms. Apart from these critical additions, Bard’s user interface has also been upgraded, with users being able to pin older conversations as well as being able to continue previous chats. These upgrades have worked to present Google Bard as a refined and capable chatbot that can compete with other major counterparts including ChatGPT and Claude—which has also recently garnered a successor under the moniker “Claude 2.” All of these improvements serve to reinforce Google’s position in the tech market as a leading producer of AI tools.
Google Bard AI’s Future Plans
Google plans on enhancing its other offerings by using its learnings from its current AI ventures like Bard.
Given that Google controls and contributes to a large market, the possibilities with Bard and its future upgrades are plenty. Google is focusing its abilities on ensuring better security and preventing jailbreaks while also ensuring its chatbot remains competitive in the market. Enhancing responsible AI features, Google also seeks to actively reduce harmful and false responses from its AI offerings by cutting down the chances of hallucination in its chatbots. Unconfirmed rumors suggest that Google might also be looking to introduce Bard into its existing Android framework, which it uses to support millions of smartphone users across the world. Given that ChatGPT already has an iOS and Android application, Google, too, might be looking to optimize its chatbot for direct usage through handheld devices. Regardless, Google’s metrics indicate that Bard’s popularity is on a steady rise and will continue to remain an important player in the chatbot market.
FAQs
1. How many languages does the new Google Bard support?
Google Bard supports 40 different languages including Mandarin, French, German, Japanese, Korean, and Hindi, among others.
2. How many countries is Google Bard available in?
Following major accessibility and location updates, Google Bard is available in over 180 nations across the world, making it one of the most widely distributed AI chatbots in the market.
3. Can Google Bard write code?
Yes, Google Bard can now write and execute code following recent updates that moved the chatbot to the better PaLM 2 language model. Since the new LLM is trained on numerous sources containing open-source code, it has grown quite capable of writing code to respond to user queries and requests.