Lifeboat Foundation

Right now it’s easiest to think about an artificial intelligence algorithm as a specific tool, like a hammer. A hammer is really good at hitting things, but when you need a saw to cut something in half, it’s back to the toolbox. Need a face recognized? Train an facial recognition algorithm, but don’t ask it to recognize cows.

Alphabet’s AI research arm, DeepMind, is trying to change that idea with a new algorithm that can learn more than one skill. Having algorithms that can learn multiple skills could make it far easier to add new languages to translators, remove bias from image recognition systems, or even have algorithms use existing knowledge to solve new complex problems. The research published in Proceedings of the National Academy of Sciences this week is preliminary, as it only tests the algorithm on playing different Atari games, but this research shows multi-purpose algorithms are actually possible.

The problem DeepMind’s research tackles is called “catastrophic forgetting,” the company writes. If you train an algorithm to recognize faces and then try to train it again to recognize cows, it will forget faces to make room for all the cow-knowledge. Modern artificial neural networks use millions of mathematic equations to calculate patterns in data, which could be the pixels that make a face or the series of words that make a sentence. These equations are connected in various ways, and are so dependent on some equations that they’ll begin to fail when even slightly tweaked for a different task. DeepMind’s new algorithm identifies and protects the equations most important for carrying out the original task, while letting the less-important ones be overwritten.

Continue reading “Artificial intelligence has a multitasking problem, and DeepMind might have a solution” »

Adam Savage gets up close with the one-of-a-kind 3D-printed endoskeleton Weta Workshop made for the upcoming Ghost in the Shell. Chatting with Weta Workshop technician Jared Haley in the studio’s 3D modeling room, Adam learns about the experimentation and prototyping necessary to make this gobsmackingly beautiful prop, which is made of several hundred individual pieces!

Shot and edited by Joey Fameli

Continue reading “How Weta Workshop Made Ghost in the Shell’s Robot Skeleton!” »

Massive and complete automation could enable industrializtion of the moon and space. By using some larger human colonies along with the robots then it would be more robust and less dependent on perfect automation.

Advances in robotics and additive manufacturing have become game-changing for the prospects of space industry. It has become feasible to bootstrap a self-sustaining, self-expanding industry at reasonably low cost. Simple modeling was developed to identify the main parameters of successful bootstrapping. This indicates that bootstrapping can be achieved with as little as 12 metric tons (MT) landed on the Moon during a period of about 20 years. The equipment will be teleoperated and then transitioned to full autonomy so the industry can spread to the asteroid belt and beyond. The strategy begins with a sub-replicating system and evolves it toward full self-sustainability (full closure) via an in situ technology spiral. The industry grows exponentially due to the free real estate, energy, and material resources of space. The mass of industrial assets at the end of bootstrapping will be 156 MT with 60 humanoid robots, or as high as 40,000MT with as many as 100,000 humanoid robots if faster manufacturing is supported by launching a total of 41 MT to the Moon. Within another few decades with no further investment, it can have millions of times the industrial capacity of the United States.

Welcome to the world’s first robot-operated 3D printer factory.

This is about as futuristic as it gets.

This could be the battle of the century… 😂 😂.

The chief executive of listed accounting software business Xero has claimed machine learning-based automation will be a bigger change than the advent of cloud computing, as it starts to offer options to automate accounting tasks.

Xero boss Rod Drury said the company would unveil a new feature to its software this week which would automate the coding of invoices and bank transactions for its small business customers, work that has been conducted personally by business owners or accountants until now.

The process was targeted for automation after Xero’s Find & Recode feature showed 3.1 million invoices had been incorrectly recorded by its 862,000 subscribers in the 18 months to September 2016. It is the first introduction of machine learning automation at Xero since it shifted its infrastructure to run on Amazon Web Services in 2016.

Continue reading “‘Bigger than the cloud’ … Xero makes first move towards automated accounting” »

Since appointing the former Microsoft executive Lu Qi as the new Chief Operating Officer and Vice Chairman, Baidu has reshuffled its senior management team and company structure to focus more on artificial intelligence.

The self-made billionaire sees tremendous opportunity in machine learning.

AI and machine learning technology is ready, but simply imposing it on an existing organization without preparing your people and processes will not deliver the best results.