Google and Alphabet CEO Sundar Pichai speaking at Google I/O 2024. Image: Google
Gemini continues to evolve and expand across Google’s ecosystem, more AI features are coming to Search and users will soon be able to create their own customised AI models.
Google continues to bank on the “Gemini era”, with AI-powered services being the core theme of the tech giant’s annual I/O event.
The focus on AI is unsurprising – Google has been going all out over the past year and a half to catch up with Microsoft, its key rival that gained an AI edge thanks to its partnership with OpenAI. Last year’s I/O event had Google “reimagining” all of its core products with AI.
At the latest developer conference, the references to AI throughout the keynote were almost overwhelming – the word AI was said roughly 120 times, according to Google itself.
“With each platform shift, we have delivered breakthroughs to help answer your questions better,” Google and Alphabet CEO Sundar Pichai said during the event. “Google Search is generative AI at the scale of human curiosity. And it’s our most exciting chapter of Search yet.”
Let’s look at some of the key updates Google has revealed during its I/O event.
AI Overviews for Google Search
One of the most interesting – and fast approaching – announcements was a way to simplify and speed up online searches with AI. To this end, Google announced the roll-out of AI Overviews, a feature from the company’s experimental Search Labs.
This feature creates an AI-boosted answer for a user’s Search query, with generated summaries, tips and links to referenced sites. Google plans to bring this feature to “everyone in the US” this week and expand it to more countries over time. The goal is to have the feature available to more than 1bn people by the end of 2024, according to Google VP and head of Google Search Liz Reid.
“With AI Overviews, people are visiting a greater diversity of websites for help with more complex questions,” Reid said. “And we see that the links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query.
“As we expand this experience, we’ll continue to focus on sending valuable traffic to publishers and creators.”
The Gemini era is spreading
Gemini is Google’s flagship AI model, the evolution of Bard that aims to challenge OpenAI and its evolving GPT creations.
At the I/O event, Google announced a number of upgrades to bring Gemini models to more users. The company said Gemini 1.5 Flash and 1.5 Pro are now available in public preview for users in more than 200 countries and territories.
Gemini 1.5 Pro is the company’s “cutting-edge model” that is available for Gemini Advanced subscribers. The company claims this model has a number of technical advances including a greatly expanded context window starting at 1m tokens.
Sissie Hsiao, Google VP and general manager of Gemini experiences and Google assistant, claimed this context window is “the longest of any widely available consumer chatbot in the world”.
“A context window this long means Gemini Advanced can make sense of multiple large documents, up to 1,500-pages total, or summarise 100 emails,” Hsiao said. “Soon it will be able to handle an hour of video content or codebases with more than 30,000 lines.”
Google also revealed that Gemini is being upgraded on mobile, as Gemini in Android Studio will support “multimodal inputs” later this year using Gemini 1.5 Pro. Gemini Nano – the company’s model for efficiently running tasks directly on mobile devices – is also being expanded to more devices later this year.
Personalised Gemini chatbots
The use cases of AI models are massive, so Google is letting some users create their own customised versions of Gemini.
Subscribers of Gemini Advanced will soon be able to create Gems – customised chatbots designed for specific purposes such as a gym assistant, a coding partner, a creative writing guide and more.
Google also said these Gems can be quickly created through simple text prompts, letting users simply describe what they want their personalised Gemini to be before the model creates a custom bot based on those instructions.
A beta boost for AI developers
Google also launched a more detailed, open-source framework for developers looking to build their own AI models.
This beta framework is called Firebase Genkit and is built for JavaScript and TypeScript developers. Google said this kit will help developers integrate AI-powered features into their new or existing apps.
“Genkit lets you integrate diverse data sources, models, cloud services, agents and more with the code-centric approach you are already familiar with,” Google said in a Firebase blogpost.
A push into AI music
Google is also making a push into the AI-generated music space, according to a post by Google DeepMind.
DeepMind said it has been working with YouTube to create the Music AI Sandbox – a “suite of AI tools to transform how music can be created”.
“To help us design and test them, we’ve been working closely with musicians, songwriters and producers,” DeepMind said. “These tools allow users to create new instrumental sections from scratch, transfer styles between tracks, and much more.”
YouTube shared a small batch of music demos created using this AI Sandbox to showcase its capabilities. But AI-generated music has been criticised by some artists in the industry.
Find out how emerging tech trends are transforming tomorrow with our new podcast, Future Human: The Series. Listen now on Spotify, on Apple or wherever you get your podcasts.
