Machine learning, a type of artificial intelligence that employs software to interpret and make predictions from large sets of data, is in popular demand in Silicon Valley. Some of the largest of those companies such as Microsoft, Facebook, and Apple have thrown their hat into the ring. But it was Google that started the trend, and in order to remain innovative, Google needed to keep looking like the cutting-edge leader.
Hence TensorFlow, a machine-learning system that Google has used internally for a few years. Today, Google is taking it open source, releasing the software parameters to fellow engineers, academics and hacks with enough coding skills. There is no denying that learning systems have made it possible to create and improve apps when it comes to speech and image recognition technologies.
For example, Google Photos have benefitted from their own machine learning system, called DisBelief. Developed in 2011, DisBelief has helped Google build large neural networks, but it has its limitations, including difficult configurations and its inability to share code externally. As a result, the company has open-sourced TensorFlow, which was designed to fix the shortcomings of DisBelief. However, it’s important to note that it only allows for part of the AI engine to be open-sourced.
By releasing TensorFlow, Google aims to make the software it built to develop and run its own AI systems a part of the standard toolset used by researchers. It may also help Google identify potential talent for the future.
Comcast has recently offered Atlanta customers the option of getting an unlimited data plan for an extra $35 a month, as they have been subject to the data cap (300 gigabytes) since 2013. It’s a limited trial, but the economics of Comcast’s unlimited plan make it a potentially dramatic shift in the way we’ll buy broadband in the future. If Comcast decides this trial is a working model, other markets where the data cap is in place could start seeing similar offers, and it’s not crazy to think the plan might someday roll out nationwide.
Cellular providers have been selling mobile Internet like this for years. You know how it works: Go over your monthly limit, and you’ll face penalties such as reduced speeds or expensive overage fees. In the select markets where Comcast has experimented with data caps, the company assesses a $10 charge for every 50 GB a customer uses beyond their monthly limit. And yes, some Comcast users actually reach this point, paying as much as $30 a month in overage fees. When Comcast surveyed these folks, it found that 60% were willing to pay a flat $30-$40 a month extra to be freed from overage payments, hence the $35 a month that Atlanta customers pay.
You can probably see where this is headed. Even if you don’t use a ton of data now, more of our work and play is moving to the Web. Netflix is the cause of one-third of the country’s Internet traffic, and we’re only just getting started with driverless and connected cars, smart appliances and other devices associated with the Internet of Things.
However, Comcast isn’t the only one looking into data cap-associated fees for unlimited usage. A study last year found Internet providers everywhere could benefit financially from introducing data caps and other features associated with metered usage plans.
So how will this possible change effect mobile users? You might be more inclined to have your data increased, as it is much more convenient to have your data readily available to you on your mobile. How will these data limits persuade you to forget home internet all together, and opt for mobile data instead?
Not every e-mail that you get in your inbox deserves a well thought out reply, which is why Google is using the power of machine learning to make email triage a little bit faster.
Google announced the new “Smart Replies” feature for its Inbox email client, which gives users of its service up to three quick options to send back in reply to emails based on a machine learning analysis of the message’s content. People can use the short replies as either a way to quickly respond, or a way to start a longer message. This could prove to be especially useful for mobile users who would rather reply with a quick response, rather than type out a whole e-mail on their phone. This is the latest example of Google’s effort to teach machines how to take over some of the tasks typically handled by humans.
The new feature is available to all consumers who use the free version of Inbox, as well as the more than 2 million businesses who pay for Google’s suite of applications designed for work.
Facebook is planning on tweaking its policy that requires members of their social media site to use their real or “authentic” names on profiles, as advocacy groups such as the Electronic Frontier Foundation and American Civil Liberties Union have shown discontent in the requirement.
The strictly enforced “real names” policy requires users to use the names that friends and family know them by, as Facebook says it helps to root out online bullying and makes users more accountable. However, the policy has seen many users suspended from Facebook despite using authentic names, with online trolls taking advantage of it to report sections of users. Transgender individuals who have chosen a new name to match the gender they identify with say they have been affected by the policy, as have drag queens and Native Americans.
Facebook has said they would add new tools that improve how users confirm their name on Facebook when signing up, and make it more difficult for trolls to target individuals. When users are asked to confirm their name (which it can do when users are reported or when a moderator questions an account) they will be allowed to add additional details to provide context. Secondly, people that want to report profiles for using non-authentic names will have to provide additional information about why they are reporting an account.
These new changes are part of the latest in a series of tweaks to Facebook’s real names policy, which include it demanding “authentic” rather than “real” names, and allowing users to verify names using more methods than just government IDs.
Maybe you slouch in your chair. Maybe you have tried to make using a standing desk a habit. Maybe, your preferred working position is curled up on a sofa. The California startup Altwork has what may be the solution: the Altwork Station.
While adjustable sit/stand desks have been done before, the Altwork Station takes things to the next level: it’s an integrated workstation combining seat, desk, and monitor stand, and it’s all electrically controlled to support not just sitting and standing but also a horizontal position: you lie back with your monitor or monitors above you. The keyboard and mouse stay affixed to your desk through the magical power of magnets.
At the push of a button, the desk can tweak the position of everything or fully shift back into sitting or standing. As the back moves, the monitor moves with your eyes, the desk moves with your hands, and the back headrest shifts slightly in or out to best support your head.
It’s available now to pre-order for a still-steep discounted price of $3,900, but once that promotional price period is over, the Altwork Station will increase to an eye-popping $5,900. The company says its trying to price things in the same realm as a really high-quality mechanically adjustable desk alongside a similarly excellent chair and a monitor arm, but no matter how you slice it that’s a lot of cash.
Ultimately, it’s a product that’s hard to recommend to most normal people, but it’s also a fascinating engineering study. It’ll be interesting to see if Altwork can find an audience with this product — and if it does, hopefully it can bring some of this technology to more people at a lower price point down the line.
On average, there are more than 3.5 billion Google searches every day, and a small percentage of these are requests that have never been made before. In an attempt to handle these obscure and hard-to-find searches, Google has developed an artificial intelligence system called RankBrain. According to Google, RankBrain tackles a “very large fraction” of all the total number of Google searches, which are about 15% of the millions of queries it receives every second.
Google searches are ranked based on ‘hundreds’ of signals including location, key words, the site’s ranking and more. Its current technology relies on discoveries and insights that people in information retrieval have had. RankBrain instead uses AI to turn words into so-called ‘vectors’ the computer can understand more easily. If it sees a word it doesn’t recognize, it takes an educated guess at what it could mean, based on other words and phrases that may have a similar meaning.
The system then filters the results and presents the most appropriate links to the person making the search. Every time it makes these guesses, it monitors how the person making the searches responds to the results and can adjust its filtering process accordingly.
Google says it’s only used RankBrain to handle this massive search load for the past few months, but Google says RankBrain is actually better at predicting top search results than their own search engineers.
The search feature on Facebook has always been good for tracking down an old friend or a page to follow, but usually Google was the better search engine for keeping up with news or finding a story from a year back. Facebook hopes to remedy that with a new, expanded search that sifts through its 2 trillion posts from all over the world — not just those from your friends and followed pages.
Now, for example, if you were to type “Mets World Series” into the search tool on Facebook, you would be hit with top post results from the Wall Street Journal and CBS Sportsline, followed by several posts made by your friends or public groups about the Mets.
The search was updated by Facebook in hopes to spark public conversations with strangers around the world about shared topics or news stories of interest. Seeing what your circle of friends has to say is one thing, but seeing what the world is saying is another.
Recently, Google announced its newest investment: a wind power project in Kenya that, when completed, will be the continent’s biggest wind farm. The agreement includes buying a 12.5% stake in Africa’s largest wind project, Kenya’s Lake Turkana, from Danish wind turbine manufacturer Vestas Wind Systems A/S.
The 310-megawatt Lake Turkana wind park is set to about 15% of Kenya’s electricity needs, based on current generation capacity. The nearly $1 billion wind project offers the scale of infrastructure that international organizations say Africa needs for the continent to unleash its vast economic potential. Annual economic growth in sub-Saharan Africa has averaged 5% in the past decade, and an increased energy production would boost growth even more.
Google so far has committed $2 billion to 22 clean energy projects, including the continent’s largest solar project in South Africa. The company sees a big opportunity in fast-growing markets with rich renewable energy resources, and the Lake Turkana project would help reduce Kenya’s reliance on fossil fuels and emergency diesel generation.
Google and Vestas have previously cooperated on the 270-megawatt Alta Wind Energy Centre in southern California and the powering of a Google data center in Finland.
The Federal Aviation Administration is getting closer to announcing rules for using recreational drones, and will likely start with the requirement to register every drone once it is purchased. However, the registration will only be able to help the FAA keep count of how many there are purchased, and to identify drones that cause collisions and break rules.
Will this signal the start of other laws to follow? Will previous owners of drones need to also register? Surely, registering the hundreds of thousands of drones already in the U.S. that are used for recreational purposes will not be an easy feat.
The number of drones has been steadily increasing over the years, as the FAA has reported pilot sightings of drones over 650 times this year. This number has jumped greatly in one year as pilots reported sighting a drone 238 times for most of 2014. An even greater influx of drones in the sky are expected shortly after the holidays as the FAA has estimated that 1 million drones are expected to be sold for Christmas.
“Autopilot” features such as steering and parking will be available for newer Tesla Model S sedans today, but CEO Elon Musk cautions that drivers should still keep ahold of the steering wheel. If the steering wheel has no hands on it during the automatic driving, a notice displaying “hold steering wheel” will illuminate on the dashboard. Also, in more difficult navigating conditions, an audio alert will come on and if that also is ignored, the car will slow and eventually stop.
Tesla, which this month unveiled its Model X SUV, has been the U.S. pioneer in luxury electric cars charged by batteries. Its expertise in software has made it a leader in self-driving features, which more traditional carmakers have been slower to develop. Musk estimated that within three years, cars will be able to drive “from your driveway to work without you touching anything,” but regulatory approval could take years. He added that regulators would need data showing that self-driving cars work.