Sarah Yáñez-Richards
New York, Oct 1 (EFE).- The tech giant Microsoft announced on Tuesday that its generative artificial intelligence (AI), Copilot, now has voice, eyes, and emotional intelligence to be a "companion" that users can turn to not only for help but also for conversation.
Mustafa Suleyman, the CEO of Microsoft AI, said today during the presentation of these new features at an event in New York that computers have now learned the language of people and that this "completely changes the interactive paradigm between humans and machines."
"These machines now understand our voice. They can perfectly transcribe what we have said. They can react to our intonation. It's at such a good level that you can have a smooth, relaxed, and informal conversation, where it picks up the emotional intelligence of your speech, and it feels like you're having a normal conversation," emphasized the co-founder of Google DeepMind.
According to Suleyman, for now, Copilot Voice, the chatbot where users can converse with Copilot, does not store conversations, nor does the company keep information to train the AI.
This AI has four voices and is being launched today in Australia, Canada, New Zealand, the UK, and the US on mobile apps iOS and Android, on the web at copilot.microsoft.com, and through the Copilot app for Windows.
According to Microsoft, Copilot Voice will initially be available in English, but during a demonstration, an EFE journalist managed to have smooth conversations with the machine in French and Spanish.
Like other AIs, its responses may be well-formed but contain false information. This is known as hallucinations, which is why users cannot "converse" with the AI on certain topics such as elections, medical queries, or legal warnings.
"Until we have very high accuracy rates and very low false positive rates, we will not extend into these other areas, but that moment will come. It's just a matter of a few years," Suleyman revealed.
The new eyes and reasoning ability of Copilot
The eyes of the AI will be Copilot Vision, a chatbot that can see the user's browser, understand both text and images, and provide advice or summaries about what it sees.
According to Microsoft, this application can be useful for organizing travel accommodations on Airbnb - as it can listen to the user's preferences and recommend a specific option based on that - or to select a movie according to reviews that critics leave on Rotten Tomatoes - as it can summarize all the comments available on the website.
This feature will not be available on all websites and is still in development. Therefore, it is only accessible in the beta version through Copilot Labs in the US.
The company also introduced Think Deeper - which is also only available in Copilot Labs - in which the AI takes more time to reason its response, which, according to the company, allows it to respond to a complex answer in which, for example, a list of pros and cons is cited.
"We designed it to be useful for all kinds of practical and everyday challenges, such as comparing two complex options side by side. Should I move to this city or that one? What type of car best suits my needs? And so on," Microsoft explained in a statement.
The tech giant also announced today Copilot Daily, an audio summary of news and weather information that Copilot reads as if it were a podcast.
"We work directly with Reuters, Financial Times, USA Today and pay for this content. But over time, it will also include emails, calendar events, and daily tasks. The idea is really to summarize only the things that (the user) needs at dawn," detailed Suleyman. EFE
(video)