Google will host its I/O developer conference today and the company will make artificial intelligence (AI) the focus of the event. That is hardly a surprise considering the mainstreaming of AI we have seen in recent months. Microsoft – through its partnership with OpenAI – has been leading that but Google is now in a position to start rivaling Microsoft.
It is well known the company was initially caught off guard by how quickly Microsoft was able to adjust its products to integrate AI such as GPT-4 and ChatGPT from OpenAI. I/O will show what Google has done to catch up. According to CNBC, the company will unveil its new large language model LLM at Google I/O.
Known as PaLM 2, the new AI chatbot will support more than 100 languages. CNBC found internal documents that show during tests Google's AI was able to perform coding, complete math tests, and analyze and generate creative writing. During testing, the company has been describing PaLM2 as a “Unified Language Model”.
Google's first PaLM language model was announced in April 2022, with an API launching for the model in March 2023. At I/O, the company will detail how the next generation of the AI will transform its own products. Furthermore, the company will also discuss its plans to further Bard, how AI will integrate in Workspace, and potentially image generating AI.
Building Bard and How it Will Underpin Google's Services
Bard is a new tool from Google that uses a streamlined and powerful version of LaMDA, a large language model (LLM) that can produce natural and coherent responses to any type of text input, such as questions, commands, or prompts.
Bard aims to help you enhance your efficiency, speed up your creativity and spark your interest. Bard is not a search engine like Microsoft's Bing Chat, but a direct interface to an LLM. However, you can easily access Google Search to verify Bard's responses or find sources across the web.
I previously reported how Google is working on bringing AI to its dominant search engine. It is likely Bard will form at least some part of this solution when it launches. Bard is currently in limited preview for users in the US and the UK.
It is similar to ChatGPT from OpenAI, but ChatGPT is more advanced and has been updated several times, now running on the GPT-3.5 AI model from OpenAI. Microsoft uses a ChatGPT integration and GPT-4 in its Bing Chat solution.
Google was caught off guard by the launch of Bing Chat and hurried to announce Bard. This decision upset employees and the situation got worse when Bard mostly malfunctioned during its presentation demo.
CEO Sundar Pichai also seemed to make odd decisions, such as requiring employees to test Bard for hours each day. In an interview, the CEO admits Bard is a prototype and will improve with time.
Chasing AI Productivity with Bard and Workspace Automation
In March, Microsoft presented an integration with GPT-4 on its own Bing and Microsoft Graph AI known as Microsoft 365 Copilot. Essentially, the company has brought LLM and generative AI capabilities to its Microsoft 365/Microsoft Office suite, including Word, PowerPoint, Excel, Microsoft Teams, Outlook, Microsoft Viva, and more.
Once again, Google was not ready to compete with Microsoft directly. However, the company has since begun testing LLM AI within its Docs and Gmail app. Since then, the AI integration has been in testing with a small number of “trusted testers” within Google. Earlier this week, Google expanded the number of users testing the features 10x.
Also this week, it emerged Google is ready to bring Bard to it Pixel devices, is working on bringing deeper chatbot AI to Google Search, and making Bard available for Google Workspace customers. While most of these are unconfirmed by Google, it seems the company is preparing a big push and many announcements at I/O later today.
Tip of the day: The Windows Sandbox gives Windows 10/11 Pro and Enterprise users a safe space to run suspicious apps without risk. In our tutorial we show you how to enable the Windows Sandbox feature.