launch of the AI ​​Seeker, Project Astra, Veo vs. Sora and the new Gemini

Google kicked off its annual developer conference this afternoon, Google I/O. This is an event whose fame has traditionally been monopolized. Android and Google’s app ecosystem, but the company already provided an important piece of space at the first AI conference last year. This year 2024 practically nothing more was said and left an important series of announcements. Many of them, as is often the case in this area, are still available in the future or very limited in the present, but there are also specific products that can already be used.

AI Reviews. AI is now available in Google Search in the US.

Google announced last I/O the emergence of artificial intelligence in the search engine, which represents a significant change to Google’s advertising-based product and business model. Over the past year he has become known as Experience generating searches in Google, SGEand was available to a limited number of users in Search laboratories. Now It launches for free in the US and in English.other countries and languages ​​will appear later.

The Google search engine does not provide ordered lists of websites so that the user can choose what seems most suitable, but rather includes a new module called AI Reviews.

something like an IA review that response to user search in natural language and with links. It uses a version of Gemini adapted for this purpose and is multimodal. That is, it can understand various input formats, not just text. Also sound and picture.

Doubt gripping Google How AI can impact your advertising business. In accordance with Liz ReedVice President of Company Search, Links displayed in AI reviews receive more clicks than those included in a traditional results list.but it’s worth asking why the user would click on the link and go to another website with more advertising. if you already have the necessary information chewed by AI and available in a search engine. Another case is search queries related to shopping or bookingwhere Reed’s statement fits best.

You will also be able answer complex, multifaceted questions. For example, you can ask about the best Pilates places in the city, how far away they are and what offers they have for new clients, and AI Reviews will provide all the necessary information in natural language, requiring minimal cognitive effort from the user. .

Project Astra

OpenAI yesterday he presented his new language model, GPT-4oand his new voice capabilities, which make him sound like an AI assistant much like the one he played. Scarlett Johansson in film Her from Spike Jonze. The product that Google introduced in the same line is Project Astrawhich is unlikely to become its commercial name when it becomes available.

Near virtual assistant with artificial intelligence which will appear as a mobile application, but not only. As explained Demis Hassabisco-founder DeepMind which is now Google’s artificial intelligence division, will be able to see what is around you, identify it and answer questions about it.

Hassabis noted that Google intends to create Project Astra: “develop universal artificial intelligence agents that can be useful in our daily lives” and can understand and react like humans. Also, “remember what you see and hear so you can understand the context and act.”

In an example shown via video conference rather than live, a person uses a mobile phone to determine what is around them, for example: the area in which it is located, focusing on the window

. The surprise was that he asked the assistant where his glasses were, and then it turned out that they were glasses with camera and integration with Project Astra. It should appear before the end of the year.

Gemini in Photos: Ask for Photos

Google integrated Twins in the application Photo. Ask for photos This is a new experimental feature coming in the coming months that will make it easier to recover lost images from your gallery.

With Ask Photos, a user can perform any search in their gallery using questions in natural language. For example, “show me the best photo of all the amusement parks I’ve visited“, so you don’t have to make the selection manually.

I see and Image 3

Google also introduced new models for converting text to image and text to video. The first is Image 3which, as the name suggests, is a development of previous models offering Basically it improves the ability to generate text in images.something that continues to be random in other models such as DALL-E And Middle of the road.

I see this is the answer to Nora, OpenAI’s video-text AI that left everyone speechless a few months ago. This AI will be able create videos with 1080p resolution and 1 minute duration Using user-supplied prompts with deep understanding of natural language to create videos that precisely match the query, you can work with visual concepts like “time lapse” or “aerial landscape photography” and highlighting moments. in physical modeling in the scene. When will they be available? We will have to wait.

Gemini 1.5 Pro with 2 million tokens, Gemini Flash

Gemini arrived last year to replace bard and this was done based on a new language model. So, the nomenclature that we have at the moment is complex. It was originally launched Gemini Nano, Pro and Ultra. Nano to run locally on mobile phones such as Pixel 8, Pro with better features and Ultra, the most advanced, comparable to GPT-4

.

Then Pro came out, but not Ultra, which became Pro 1.5 with 1 million context tokens. Tokens are translated into a certain number of words and represent the volume that the AI ​​can process in a conversation with the user, the context that it can receive. Pro has now increased up to 2 million, which means you can go much deeper. Until now it was available by subscription Gemini Advanced in Spain, but only in English. Win Spanish and can perform actions such as summarizing hundreds of user emails or managing documents of up to 1,500 pages.

On top of all this, we have to add a new member to the Gemini family, who Gemini Flash. This is a lightweight version of Gemini Pro offering Faster to operate and more cost effective for large scale operation. For now, it will be available to developers and not to the general public.