This May 14, Google opened the developer conferences, just before those of Microsoft and Apple. On the agenda: AI, AI and more AI, everywhere, all the time, for everything. The great revolution is well underway.
Google’s annual high mass, Google I/O, was held online this Tuesday, May 14, 2024, from the American giant’s headquarters in Mountain View. And, the least we could say was that his program was rather busy with, as the main subject, artificial intelligence… and nothing but AI! Those who expected to discover the next Pixel 9, Pixel Tablet and other Chromecast were disappointed.
AI occupied so much space during this conference that Sundar Pichai, the boss of Google and master of ceremonies for the occasion, concluded the event by calculating the number of times these two initials had been pronounced: 120. In two hours, it’s a good average which also sets the tone for the American firm’s main concerns.
In the absence of hardware, it is therefore on the software side that all of Google’s efforts seem to be focused. At the heart of this galaxy of new services is Gemini, the multimodal artificial intelligence model (understand that it can tackle not only text but also photos, videos and sound). After a timid and somewhat chaotic start (Bard’s setbacks are not entirely forgotten) Google is showing its muscles to get back into the competition and hold its own at ChatGPT 4o presented the day before by its great rival in the field, OpenAI . Here is a summary of the main announcements from this Google I/O 2024 opening conference.
AI Overviews: generative AI in the search engine
With artificial intelligence, Google wants to fundamentally reshape what has been its core business for more than 25 years: its search engine. And it is certain that the tool takes a real leap forward with the main objective: to do the research for you. Until now, by performing a query in the search box, you obtained a list of links to websites related to the subject. With AI Overviews, Gemini takes control. The top of the results page is adorned with a summary generated by Gemini that answers the requested topic. It can be supplemented with AI suggestions to refine the search as well as Maps or YouTube videos. As you will have understood, the links to the websites are relegated much further down the page. The principle allows, according to Google, to save time by avoiding opening links all the time to websites that are well referenced but which do not necessarily answer the question asked. In testing for a year in the United States, AI Overviews will first be deployed on American soil and then opened to other countries around the world. Google hopes to reach one billion users before the end of the year. It is not yet known whether AI Overviews complies with the DMA requirements in Europe or whether Google will have to make some adjustments on the old continent. One thing is certain, AI Overviews will radically change the way we get results with Google Search. And the firm nevertheless wishes to reassure website publishers who already imagine themselves reviewing their SEO from top to bottom: ” As we expand this experience, we will continue to focus on sending valuable traffic to publishers and creators. As always, ads will continue to appear in dedicated locations throughout the page, with clear labeling to distinguish organic from sponsored results “.
Google Search: conducting queries with a video
New function also coming, the possibility of using a video instead of text to conduct a search. The example used for the demonstration was based on the malfunction of a vinyl turntable. The AI analyzes the sequence frame by frame, extracts all possible information (such as the make and model of the device) and “understands” the problem. It then delivers results in summary form (the websites used as references are still not listed first) to solve the problem. This feature is still in testing in the United States at the moment and will be expanded to other regions over time according to Google.
Google Workspace: Gemini invites itself to meetings
Workspace brings together Google’s suite of collaborative work tools. And soon, companies will be able to count on a new virtual employee embodied by Gemini. He will have his own account and will be responsible, for example, for finding documents related to the same project, producing a summary, checking schedules, developing dashboards, etc. In short, Gemini will be a model employee, available day after day. and night 7 days a week.
Google Photos uses AI
Google Photos, the in-house image manager, has been around for nine years now. Google says that 6 billion photos are sent to its servers daily. The service already uses AI for image editing. But Gemini is now here to organize the photo library. We can then ask him to find anything and everything. The most stunning demonstration during the conference showed how with the query: “what is the registration of my car?” Google Photos managed after analysis to identify a photo, among thousands stored in the photo library, presenting the plate number of the good car by comparison. Another possibility is to ask to create an album representing the evolution of an activity over time. The example used to illustrate the concept was the progress in swimming of a little girl, from the first learning sessions to competitions. Here again the goal is to save time to avoid having to rummage through thousands of photos taken over several years.
Gemini installs in Android 15
The next version of Google’s mobile operating system, which generally hogs a large part of this high mass, has barely been mentioned. And obviously, it is through the prism of AI and in particular Gemini that Android 15 was discussed. We must prepare to say goodbye to Google Assistant which will be replaced by Gemini. The AI here will take the form of a floating window which is superimposed on the previously opened app. It will be possible to use content generated by Gemini by dragging it into the app of your choice. In addition, the AI will also watch your back, particularly with fraudulent calls. She will report the risk of scam during the call.
Project Astra: Google’s response to ChatGPT
OpenAI, the originator of ChatGPT, presented ChatGPT 4o the day before, a major evolution of its language model with impressive interaction capabilities combining text, voice and images. Google’s response was not long in coming and is revealed in the form of the Astra project. As its name suggests, the process is still in the development stage but the video presented during the conference suggests immense possibilities (if it is not again a somewhat misleading demonstration ). Using a smartphone pointing its camera at various objects in video, the AI is able to analyze the environment, solve problems, interact in real time with the user, etc. We even detect in the video a pair of Google Glass-style glasses which were a flop when they were released in 2011 but which could well rise from the ashes with an on-board AI such as Gemini.
Generative AI for creating images, videos and sounds
Generative AI, which allows you to create content in the blink of an eye, also represents a major pillar at Google. The firm thus took advantage of the developers’ conference to present Imagen 3, capable of generating even more photorealistic images by knowing how to interpret longer prompts (textual descriptions). Also there, Veo, which aims to be a direct competitor to OpenAI’s Sora. The goal here is to generate full 1080p videos from written or even voice prompts. It is possible to indicate the nature of the shots to be generated (traveling, aerial, panoramic, etc.), the lighting ambiance and many other aspects.
Finally, musicians also benefit from Music AI Sandbox, which allows you to generate loops or enrich pieces already created using generative AI.