Transcription
What if a voice assistant could communicate with us as easily and naturally as a good friend? It seems Google has taken a huge step in exactly this direction. They rolled out a big update for their E Gi. And today we'll look at how this can completely change our communication with technology. Well, I think almost everyone has encountered this, right? When the assistant doesn't understand you immediately, answers something completely wrong, or just sounds, well, like a robot, it can be very annoying, to be honest. But it seems this era is slowly coming to an end. Google has seriously taken on this problem. And the new update for Gemini is their answer. And we're not talking about some minor improvements here, no. There are two real breakthroughs. The creation of truly live voice agents and a translation technology that seems to have come straight out of science fiction. Let's see how it all works. So, let's start with the first, with live voice agents. And immediately forget about simple commands like "Turn on the light." Here we are talking about artificial intelligence that is capable of conducting complex multi-stage dialogues and helping with complex tasks. So what does this mean in practice? Well, imagine, instead of just asking to play music, you can tell the assistant: "Help me return this item." And it will find the order number itself, call customer support itself, and conduct this whole, sometimes tedious, dialogue for you. This is no longer just an assistant, it is, in essence, a personal assistant. How did they manage it? It's all due to three key improvements. Firstly, and now much more accurately, it accesses external tools. That is, it can, say, book a ticket or check an order status directly in real-time. Secondly, it follows instructions much more reliably, even if they are confusing. And thirdly, it better keeps the context of the conversation in mind, so you won't have to repeat the same thing five times. And here's a figure that best shows this leap. The accuracy of following instructions is a whole 90%. In practice, this means that in nine out of ten cases, the AI will do exactly what it was asked to do, without errors or misunderstandings. And that, you know, is a huge step towards truly reliable assistants. And here's a graph that shows how well Gemini handles complex tasks compared to previous versions and even competitors. On a special test that simulates real requests, the new model shows a result of 71.5%. And this is confident leadership. So this is not just words, but a proven fact. But, you know, the success is best told by those who are already using the technology. For example, Shopify is a giant in the world of online retail. They have already implemented these capabilities, and their feedback is simply amazing. Customers forget they are talking to AI within a minute. In my opinion, this is the best compliment for a technology that strives to be human-like. But natural conversation is only half the battle. What if this conversation could overcome absolutely any language barrier? This is where we move on to the second, and perhaps the most futuristic, part of the entire announcement, real-time speech translation. Just imagine two seemingly fantastic situations. First: you are walking down the street in another country, and you hear all the signs and conversations of passers-by in your headphones in your native language. This is the continuous listening mode. And second: you sit down at a table with a foreign partner and have a live dialogue, and the translation happens instantly. This is already a two-way conversation. And the process itself is ingeniously simple. You speak in your language. And the phone immediately speaks the translation for the other person. They reply in their language, and the translation is heard in your headphones. The delay is minimal, and this creates the feeling of an absolutely natural, seamless dialogue. It's just magic. But it's not just about speed. This technology is incredibly smart. It doesn't just translate words, it preserves intonation and speech tempo, conveying even the speaker's emotions. It understands over 70 languages, automatically detects which language is currently being spoken, and can even filter out street noise so that the interlocutor is clearly heard. All of this together creates the most natural translation experience ever. All of this sounds great, of course, but the main question is: how can you try it? Let's move on to the last part and figure out how to access these new capabilities right now. It's simple. There are two options. Regular users can try the beta version of the translation feature directly in the Google Translate app on Android. Developers, on the other hand, can already start integrating all these voice features into their products through the Vertex AI platform and Gemini API. An important clarification at this moment. The beta version of the translator is currently available on Android and only in the USA, Mexico, and India, but Google promises to expand the geography soon and add iOS support, so it's worth keeping an eye on updates. You know, all these technologies are more than just convenience. They literally erase the language barriers that have divided people for millennia. This opens up incredible opportunities for everything from travel to global business. And it makes you wonder, when we can perfectly understand each other, regardless of language, what new ideas and discoveries await humanity. Our analysis for today is concluded. Thank you for your attention.