10 Takeaways From the Google I/O 2024 Keynote

As LeanCode, the official consultants for Flutter and open-source contributors, we were invited by Google's team, together with Mateusz Wojtczak and Kuba Wojtczak, to take part in the Google I/O 2024 event. This is always one of the most awaited tech events of the year, and we were excited to be here in person.

Announcements at Google I/O 2024 have all been about and around AI*. We had a chance to witness those announcements in person in the sunny amphitheater at Mountain View, California, USA. Let me guide you through the key news, what they could mean for your product and productivity, and why I believe Google has all it takes to play a dominant role in the AI industry.

Two new Gemini services have been announced

The keynote opened with a very strong statement. Sundar Pichai said that Gemini will be gradually integrated into the tools used by 2 billion people worldwide this year. This means that Gemini will be shipped into all products from the vast portfolio of everyday-use applications and will impact the lives of billions. They don't need to have new downloads. All it takes is to update the existing apps.

From the technical perspective, two new services have been announced:

- Gemini 1.5 PRO: A powerful multi-modal AI model now available globally, capable of handling vast amounts of data and understanding various formats like text, images, and graphs; it attracted more than 1M users in the closed beta tests over the last 3 months, now this is shipped globally to 35 countries with the expanded context into 2M tokens; this will cost 3.5 USD per 1M tokens;

- Gemini 1.5 Flash: A lighter, more efficient model optimized for speed and low latency; it costs 0.35 USD per 1M tokens.

Why are tokens important? They help you to build context around your prompts to specify your intention better and provide additional sources, documents, and pdf of around 1.5K pages.

How will new Gemini services change the world? Here are 10 mindblowing takeaways.

1. Learning. JoshWoodward talked about how Notebook LM will be able to generate a study guide, FAQs, and Quizzes based on the uploaded books and study materials. He even showed the case of a conversational approach to explaining the content, where AI leads the discussion, and students can jump in to ask further questions.

2. AI Agents, aka Project Astra. Demis Hassabis showcased a universal AI agent that can be helpful in daily AI. Think about a personal assistant who can find your glasses, understand your email or the code you write, help interpret the data, be creative when inventing the dog's name, and many others.

3. Imagen3 — introduced by Doug Eck, is a new photorealistic image generation model based on the 6 billion photos uploaded to Google Photos daily.

4. VEO — is a video-generating model that will help extend images and is a direct answer to Sora.

5. Music generation — gives the ability to generate tracks based on beats and creatively transform them.

6. Google AI Search. Liz Reid introduced groundbreaking changes, some of which are already available in the US, like AI overview, multi-step reasoning, planning in search, AI sorted results, and ASK with video. Search for them in Google Labs.

7. Workspaces. Aparna Pappu announced the final boost to productivity, telling us about the new side panel that will arrive in Gmail, Google Meet, Calendar, and other favorite productivity apps from Google.

8. AI Teammate. Tony Vincent talked about the future of AI Assistants living in Gmail chat to help you answer complex questions related to project progress, etc.

9. Gemini App - a mobile multimodal application using text, images and video for prompting.

10. Android devices. Last but not least, Sameet Samat and later Dave Bruce introduced Androids powered by Gemini with on-device AI processing to give you the superpowers of Gemini, which understand the current context from the screen/app that is open at the moment so that you can ask direct questions about the YouTube video, email in your inbox, or an image in your photos.

... and many other news related to Gemma, PaliGemma, Red Teaming, Trillium, and partnership with NVIDIA.

Summing up

Google has the infrastructure, data, and products to boost the productivity of billions of people and dominate the GenAI revolution inspired by LLMs. The future looks more interesting than ever. Google I/O 2024 wasn't just a showcase of cutting-edge technology; it was a glimpse into a future where AI is not just powerful but also helpful, insightful, and accessible to all. The journey towards this future has begun, and Google is leading the way.

Attending the Google I/O 2024 event was also a great occasion to network and meet many people that are responsible for developing the Flutter framework daily.

*Btw. Fun fact: "AI," as counted by AI, appeared 121 times during the opening keynote.

In the photos: Łukasz Kosman, Mateusz Wojtczak, Jakub Wojtczak, Craig Labenz, Eric Seidel.