Google I/O 2024: recap and key highlights from Google for this year

The annual Google I/O 2024 developer conference took place this Tuesday, May 14th. In this conference, the company announces the main updates they have launched or will launch in their applications, from Google Search to Gmail or Google Photos.

As expected, Generative AI was the highlight of the Google I/O 2024 conference. Among the main new features related to Generative AI that the tech giant announced are:

Google announced Gemini 1.5 Flash and 1.5 Pro, thenew version of Gemini, its Generative AI model, a new version which is said to be faster and more efficient and is available in more than 35 languages. Both 1.5 Pro and 1.5 Flash are available in public preview with a context window of one million tokens in Vertex AIandAI Studio – currently, way larger than any other.

Additionally, they also announced the integration of this 1.5 Pro version into Google products, includingGemini Advanced andWorkspace apps.

Specifically, with this new model, they indicate that Gemini Advanced will be able to assist in tasks such as:

• As a data analyst: Revealing trends and creating custom visualizations on the fly from uploaded data files like spreadsheets.
• Interpreting images: Gemini is natively multimodal, and the 1.5 Pro model brings significant improvements in image understanding. For example, you can take a photo of a dish at your favorite restaurant and ask for the recipe or capture a math problem and get step-by-step instructions to solve it. All from a single image.

Gemma is a set of lightweight, generative open AI models. Models of Gemma are available to run on applications and hardware, on mobile devices, or hosted services. Users can use Gemma models for code generation use cases, but also for other functionalities, such as improving performance in specific tasks. In conclusion, the goal of a fine-tuned Gemma model, according to Google, is to help make generative AI solutions more targeted and efficient.

At Google I/O 2024, Google introduced the latest version of Gemma with Gemma 2, and an expansion with PaliGemma.

PaliGemma is an optimized model for image interpretation. Specifically, it is designed to offer precise performance for vision-language tasks. This includes functionalities such as captioning images and short videos, answering visual questions, understanding text in images, or detecting and segmenting objects.

Google announced that it would open-source SynthID, its watermarking tool capable of determining whether an image has been generated by artificial intelligence.

Although updates in Generative AI were the highlights of Google I/O 2024, the company also presented other notable news.

This includes updates in mobile and Android development, where Google announced that it has enhanced mobile development experiences for Android through advanced APIs, tools, and guides that improve developer productivity. Notably, Gemini in Android Studio will soon support multimodal inputs using Gemini 1.5 Pro, promising to further revolutionize Android development.

Finally, Google also introduced Project IDX in open beta to offer a simplified development experience in full-stack and cross-platform applications, integrating tools like Chrome DevTools and Cloud Run. Additionally, Firebase has evolved to support modern AI-powered applications, allowing connection to PostgreSQL databases and facilitating secure, scalable deployments from GitHub. Flutter and Dart tools have also been improved to offer significantly better graphics and application performance, along with new experimental features to optimize the development experience in Dart.