Download: Define open source AI, and replace Siri


Suddenly, “open source” is the new buzzword in AI circles. Meta has promised to make artificial intelligence open source. And Elon Musk is suing OpenAI for lack of open source AI models.

Meanwhile, more and more technology leaders and companies are positioning themselves as open source experts.

But there is a big problem—no one can agree on the definition of “open-source AI”. In theory, it promises a future where everyone can participate in the development of technology. This can accelerate innovation, promote transparency, and give users control over systems that will soon change many aspects of our lives.

But what exactly is it? What makes an AI model open source, and what holds it back? Whatever the answers, they could have far-reaching consequences. Read the full article.

— Ed Gent

Apple researchers are exploring ditching the voice of “Siri” and listening with AI instead

Story: Researchers from Apple are investigating whether it’s possible to use artificial intelligence to recognize a user when speaking to a device like the iPhone, thereby removing the need for technology to trigger words like “Siri,” according to a new paper.

How they did it: Researchers trained a large-scale language using cell phone recordings and acoustics from background noise in order to determine the behavior they would show when they needed help from the device. The results were encouraging—the model, which was built in part with OpenAI’s GPT-2 model, was able to predict more accurately than audio alone or text alone, and it grew exponentially as the sample size grew. .

Why it matters: The paper is one of several recent developments in which Apple, which seems to be lagging behind other tech giants such as Amazon, Google, and Facebook in the artificial intelligence race, plans to incorporate more AI into its products. Read the full article.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *