Toggle light / dark theme

Neural networks are powerful things, but they need a lot of juice. Engineers at MIT have now developed a new chip that cuts neural nets’ power consumption by up to 95 percent, potentially allowing them to run on battery-powered mobile devices.

Smartphones these days are getting truly smart, with ever more AI-powered services like digital assistants and real-time translation. But typically the neural nets crunching the data for these services are in the cloud, with data from smartphones ferried back and forth.

That’s not ideal, as it requires a lot of communication bandwidth and means potentially sensitive data is being transmitted and stored on servers outside the user’s control. But the huge amounts of energy needed to power the GPUs neural networks run on make it impractical to implement them in devices that run on limited battery power.

Read more

We’re working to make AI accessible by providing lessons, tutorials and hands-on exercises for people at all experience levels. Filter the resources below to start learning, building and problem-solving.

Read more

Tesla Inc. Chief Executive Elon Musk took to Twitter this week to decry the slow and costly pace of U.S. infrastructure projects, and thus a challenge — or boast — was born: The Boring Co. could build a tunnel across the San Francisco Bay far cheaper and far faster than current available estimates.

Musk was pointing to a San Francisco Chronicle editorial favoring a new Transbay Tube for public transportation over another cross-bay bridge to alleviate chronic traffic woes and public-transportation overcrowding in the Bay Area. The editorial cited a starting price of $12 billion for the new tunnel.

Probably about a tenth of the cost and a fifth of the time— Elon Musk (@elonmusk) February 28, 2018

Read more

Temperature changes large and small are happening around us all the time, and scientists have come up with a machine that can convert those fluctuations into electricity, potentially powering sensors and communication devices almost out of thin air.

The energy harvesting is done through what’s called a thermal resonator: a device that captures heat on one side and radiates it over to the other. As both sides try and reach equilibrium, the energy can be caught using the process of thermoelectrics.

According to the team from the Massachusetts Institute of Technology, the new thermal resonator could keep remote sensors or any off-grid devices powered up for years, just by using temperature swings – like the natural ones between night and day, for instance.

Read more

GALAPAGOS ISLANDS, Ecuador (AP) — It’s the biggest shark — and the biggest fish — in the sea, often found roaming in warm waters around the globe with its huge mouth agape in search of dinner.

Yet despite its hulking appearance, the whale shark has only tiny, almost useless teeth and is sometimes so docile that entire boatloads of people can swim alongside the enigmatic, spotted beast. It’s also one of the least understood animals in the oceans.

Read more

This event will be webcast live from this page.

The Technology Policy Program invites you to the launch of our upcoming report, A National Machine Intelligence Strategy for the United States.

The United States is at the precipice of a defining moment in history. Over the past five years, progress in machine intelligence (MI) has greatly accelerated. From the defeat of Go champion Lee Sedol by DeepMind’s AlphaGo program to the first deployments of fully-autonomous vehicles on public roads, recent events are challenging us to re-evaluate what may soon be possible for computerized systems. MI systems have already begun to quietly pervade a growing share of businesses, governments, and individual lives around the world, and we are only just beginning to grasp the impacts that this technological revolution will have on our economy, our society, and our national security. In our paper, we outline they key elements of a comprehensive national strategy for the United States to promote the safe and responsible development of MI, and to maintain U.S. leadership in MI technology.

Read more