Google I/O in previous years focused on Android, Pixel devices and cool features such as Magic Eraser. This year, Google I/O is all about the models.
Against the backdrop of a never ending stream of generative AI announcements and Microsoft's move to infuse its applications with OpenAI's ChatGPT, Alphabet CEO Sundar Pichai took the stage and made the case for its generative AI advances. "AI is having a very busy year," quipped Pichai. Here's the case in a nutshell.
Google has been using AI for years. Pichai said Google has been using AI to be more helpful for years with features such as smart reply in Gmail, Magic Eraser in Google Photos and various Maps innovation. "Today, we are taking the next generative AI step with a bold and responsible approach across all of our products," said Pichai.
AWS, Microsoft Azure, Google Cloud themes: Optimization, generative AI and the long game
Google is innovating with its models. The company launched PaLM 2, which has various models aimed at mobile as well as larger implementations. "25 products and features are delivered by PaLM 2 today," said Pichai.
The demos highlighted how a coder could fix a bug and add comments in Korean.
PaLM 2 was also highlighted in a medical analysis of an Xray to highlight specific use cases.
Bard is an ingredient brand. Bard has been updated and will have extensions to various applications. Built on PaLM 2, Bard will be updated to handle images and update them. A demonstration featured a natural language question about universities in Pennsylvania, refining the results through questions and putting the answer in a Google Sheet.
Bard is also out of waitlist mode and will be available in 180 countries.
Google Workspace will use generative AI to boost productivity. Generative AI capabilities are being added to Gmail, Docs, Sheets, Slides and Meet. Demos included using generative AI for job descriptions, marketing content and emails using the Help Me Write feature. The idea is that generative AI supplies the starter content for refinement. In Sheets, generative AI can create starter tables.
Search gets conversational. Google search will also use generative AI where Google will add aspects to consider when making a purchase as well as reviews from Google Shopping Graph. Google will be rolling out new experiences in search including "converse" or "as a follow up question." There will also be a conversational mode for search that aims to understand your intent. Experiences will be integrated so more results can be found with a scroll.
Constellation Research analyst Holger Mueller said: "Google elevates the search battle by making search multimodal - AI adjacent to traditional search results... and even in dialogue mode - traditional search results will show up. That's a good compromise for traditional usage - and a more fail safe approach to introducing AI into search than the current solely conversational UX approaches."
Android 14 will get generative AI features. Messages will get Magic Compose, which will use generative AI to adjust tone such as more professional or playful. There will be new customizations for wallpapers will use generative AI for cinematic photos as well as original wallpapers. Pixel devices will also leverage generative AI for Pixel Call Assist, Pixel Speech and new camera features.
More models are coming. Google outlined Gemini, a new model in development. Gemini will be able to find synthetically generated content via watermarking and metadata.