The Google I/O 2024 had many announcements and updates about various Google products and services, mainly focusing on AI improvements.

Google's new AI announcements:

Astra

Astra is a new way to interact with the world using Google's AI. It is basically a "multi-modal" assistant which can interact with you through audio, video, text etc. It can see what you can see and help you make sense of things.

VideoFX/AudioFX

Google released imagen3 which can generate the images, Veo which is an video generator tool and Google also released a AI Kitchen where you can join the waitlist.

Generative AI powered search

Google finally declared that Generative AI will also be the first class citizen of their search results page and will roll out to USA starting today. This is a big change as until so far Google had rolled it out only for few users.

But did Google nail it ?

Everyone sees OpenAI as the AI leader. Last Friday OpenAI hurriedly did a demo of their new ChatGPT-4o which was probably meant to take the thunder away from the Google's IO demos.

OpenAI has not really released their video generation tool Sora yet, neither has Google. But Google has certainly tried to show that they are not behind.

One area where Google excels and is clearly better than competitors is in terms of size of the tokens. Google's Gemini models can take upto 1M tokens as input and this number will go upto 2M tokens.

The second area where Google has beat others is their new TPU. Trillium. Cost of training is going to be a big factor for everyone and having a special purpose chip vastly benefits Google.

Need to see products

Ultimately we need to see the finished products in our hands to decide if it is really good. This remains true for OpenAI as well as Google both. Time will tell who will be the leader but right now they appear to be in the same ballpark.