Category Archives: Ai

Forbes

IBM's New PowerAI Features Again Demonstrate Enterprise AI Leadership Forbes Last November I wrote a column about IBM's newly announced AI software toolkit, PowerAI. PowerAI fulfills a special niche part of IBM's AI investmentaimed towards enterprises who want something in between Watson's turnkey solution (easy button on ...

More:

IBM's New PowerAI Features Again Demonstrate Enterprise AI Leadership - Forbes

The cognitive technologies behind artificial intelligence have the potential to fundamentally reshape how the federal government operates, according to a report from the Deloitte Center for Government Insights.

The report, AI-Augmented Government, released earlier this month, says that over time, AI will spawn massive changes in the public sector, transforming how government employees get work done. Its likely to eliminate some jobs, lead to the redesign of countless others, and create entirely new professions.

One of the biggest potential changes is the cost of doing business. The report found that millions of working hours each year (out of some 4.3 billion worked total) could be freed up today by automating tasks that computers already routinely do.

At the low end of the spectrum, Deloitte estimates, automation could save 96.7 million federal hours annually, with a potential savings of $3.3 billion. At the high end, those figures jump to 1.2 billion hours and a potential annual savings of $41.1 billion.

The report provides a taxonomy of AI systems rules-based systems, machine translation, computer vision, machine learning, robotics and natural language processing and delves into their practical applications in government. How can AI reshape the federal government? Here are seven clear benefits the report spells out.

Robotics and cognitive automation allow machines to replicate human actions and judgment, allowing people to avoid manual tasks in order to do work that requires uniquely human abilities, Deloitte notes.

For example, we can automate data entry with automatic handwriting recognition, handle scheduling with planning and optimization algorithms, and use speech recognition, natural language processing, and question-answering technology to provide customer service, the report says. Such capabilities could potentially address three common pain points for government: resource constraints, paperwork burdens, and backlogs.

As an example, the report notes that the Department of Homeland Securitys Citizenship and Immigration Services created a virtual assistant, EMMA, that can respond accurately to human language.

EMMA uses its intelligence simply, showing relevant answers to questions almost a half-million questions per month right now. Learning from her own experiences, the virtual assistant gets smarter as she answers more questions. Customer feedback tells EMMA which answers helped, honing her grasp of the data in a process called supervised learning.

The report notes that cognitive automation can perform tasks at previously impractical scales, speeds, and volumes, allowing for both resource redistribution and workforce optimization.

For example, electronic document discovery locates 95 percent of relevant documents in the discovery phase of legal cases, compared to an average 50 percent for humans, and in a fraction of the time. The technology allows lawyers to sift through vastly larger document dumps, the report says. In medicine, similarly, robotic surgery aims to allow doctors to perform more operations.

A 2016Governing survey of state and local officials found that 53 percent had trouble getting their work done in a 35-to-40-hour week due to excessive paperwork burdens, the report notes.

At the federal level, Deloitte says that simply documenting and recording information consumes a half-billion staff hours each year, at a cost of more than $16 billion in wages alone. Procuring and processing information eats up another 280 million person hours, costing the federal government an additional $15 billion annually. Robotic automation could help cut that down substantially.

At the U.S. Patent and Trademark Office, the backlog of patent applications reached 558,091 in October 2015, the report notes. Patent delays can significantly hamper firms, especially start-ups; an agency study concluded that each year of delay in reviewing first patent applications that ultimately receive approval reduces a companys employment and sales growth by 21 and 28 percent, respectively, over five years, according to the report.

However, cognitive technologies can sift through large data backlogs and take appropriate action, leaving difficult cases to human experts. Robotic process automation, in turn, can reduce backlogs by performing entire end-to-end business processes on a massive scale with little human interaction.

AI technology embedded with sensors and cameras can allow agencies to track and report important information in real time, the report says.

AI-enabled real-time tracking and reporting can also enable disease surveillance, exemplifying a potentially life-saving capability, Deloitte notes. The U.S. Centers for Disease Control and Prevention has streamlined its polio virus tracking and reporting process with an AI tool that classifies virus types and separates disease reports into related clusters.

Machine learning and natural language processing can reveal patterns and guide effective responses to problems that agencies routinely track, according to the report. For example, such technologies could reveal the most vulnerable populations in public health crises or trace the origins of food-borne illness.

The Armys Medical Department is developing wearable physiological monitors that use a machine-learning algorithm to weigh the potential seriousness of wounds, to assist medics in prioritizing treatment or evacuation, the report says.

Meanwhile, the Energy Departments new self-learning weather and renewable forecasting technology, SMT, is 30 percent more accurate in solar forecasting than previous techniques (the work is being done in partnership with IBM), according to Deloitte. To improve its prediction accuracy, the system uses machine learning, information from multiple sensor networks, cloud motion physics derived from sky cameras, and satellite observations, the report says.

Like DHS, the Army devotes significant resources to answer citizens queries. The Army does so especially for potential recruits.

To help prospective recruits understand their options, visitors to the Army website encounter SGT STAR, an interactive virtual assistant that uses AI to answer questions, check users qualifications, and refer them to human recruiters.

The Army found that SGT STAR does the work of 55 recruiters, with an accuracy rate of more than 94 percent, and has increased engagement time for site visitors from 4.0 to 10.4 minutes, Deloitte notes. As of 2016, the virtual assistant had answered more than 16 million user questions.

SGT STAR uses machine learning to recognize data patterns that help it distinguish helpful answers from unhelpful ones, the report says. The more questions it answers, the more it learns and the better it gets.

The report notes that if cognitive automation and engagement are used to relieve the human worker of tasks that are rules-based, routine, repetitive, and relatively simple, humans are then free to focus on more complex, value-adding tasks.

Cognitive insights can then help federal workers perform these more difficult tasks effectively and efficiently.

Read more from the original source:

7 Ways AI Could Save the Government Money and Boost Productivity - FedTech Magazine

Viz, a San Francisco-based company looking to apply artificial intelligence to medical imaging, has raised $7.5 million in first-round funding. The round includes two tranches -- a $4.7 million round that just closed and a $2.5 million round from last fall. The other $300,000 was reserved for employees to contribute. The funding round was led byDHVC and Alphabet Chairman Eric Schmidts Innovation Endeavors. AME Cloud Ventures and Susa Ventures also contributed.

"AI in the form of deep learning is actually relatively simple when you think of it," CEO Dr. Chris Mansa told MobiHealthNews. "Its just training an algorithm to identify patterns. And doctors, when we look at medical images, we are looking for patterns that weve learned about and seen before and categorizing those patterns into diagnoses. So what we do at Viz is we use deep learning to help doctors make that diagnosis faster and get the right patient to the right doctor at the right time."

Viz is initially focusing on stroke, but intends to expand its capability to a number of health conditions where diagnostic imagery is used and time is of the essence.

"When an artery to the brain is blocked, brain cells die at a rate of 2 million every single minute," Mansa, who is trained as a neurosurgeon, explained. "With an artery to the heart [by comparison] its closer to 70,000 cells every minute. Essentially the brain is much more sensitive to schemia, or lack of oxygen, than any other part of the body. And this is why stroke is the number one cause of preventable disability in the Western world."

The company has trained a learning neural network on a large database of medical imagery from the last 10 years to create a tool that can quickly connect patients to the right specialist and give that specialist additional tools to make a diagnosis. The tool will be integrated into a hospital's existing DICOM infrastructure.

"We dont see ourselves and its not a marketing thing, its just that I know the truth about how complicated healthcare is we do not see ourselves as replacing doctors," Mansa said. "Were very much an aid and an assistant to help them do the fantastic work that they already do in a more efficient, faster way."

Viz has completed some pilots in the US and the UK and is awaiting FDA clearance. They expect to have it by late 2017 or early 2018. Mansa said that once the stroke tool is cleared, the company has a pipeline of several other products in the wings. The funding will be used to commercialize the stroke product once it's cleared and to continue to develop the future products.

Go here to see the original:

With $7.5M in funding, Viz will use AI for time-sensitive image analysis - MobiHealthNews

BBC News

Google AI defeats human Go champion BBC News "It is an incredible achievement and most experts thought an AI winning at Go was 20 years away so DeepMind is leading the field but this AI doesn't have general intelligence. It doesn't know that is playing a game and it can't make you a cup of tea ... An AI Is Now the Best Player of Humanity's Most Complicated GameGizmodo Google's AI Program Rattles Chinese Go Master as It Wins MatchNew York Times Google's Go-playing AI still undefeated with victory over world number oneThe Guardian CNNMoney -TechCrunch -Phys.Org -DeepMind all 146 news articles »

View original post here:

Google AI defeats human Go champion - BBC News

Forbes

Two Big Reasons Why Google's AI Chips Will Have A Tough Time Competing With Nvidia Forbes Not even Google will likely slow down Nvidia's artificial intelligence gravy train anytime soon. Graphics chipmaker Nvidia, whose stock has tripled in the past year, has emerged as the dominant platform for deep learning, a popular form of AI computing. Nvidia CEO Predicts AI 'Cambrian Explosion'HPCwire (blog) all 7 news articles »

Read more:

Two Big Reasons Why Google's AI Chips Will Have A Tough Time Competing With Nvidia - Forbes

Ke Jie, the world's No. 1 Go player, stares at the board during his second match against AlphaGo in Wuzhen, China, on Thursday. The 19-year-old grandmaster dropped the match in the best-of-three series against Google's artificial intelligence program. AFP/Getty Images hide caption

Ke Jie, the world's No. 1 Go player, stares at the board during his second match against AlphaGo in Wuzhen, China, on Thursday. The 19-year-old grandmaster dropped the match in the best-of-three series against Google's artificial intelligence program.

Sure, it's not the singularity (yet) but it is a rather singular achievement.

AlphaGo, an artificial intelligence program developed by Google's DeepMind lab, did not even need a third game to display its dominance over the world's best (human) Go player. On Thursday the A.I. defeated Ke Jie in Wuzhen, China, repeating its victory of two days ago and clinching a best-of-three series against the 19-year-old wunderkind.

Ke has one more chance to redeem himself, on Saturday, though if the first two matches are any indication, is chances don't look good.

That's not to cast aspersions on Ke's play. Quite the opposite, in fact. In a blog post Thursday, DeepMind called the match "a stunning work of art."

"According to AlphaGo's estimation of the match, the programme assessed the first 50 moves as virtually perfect," the post read, "and the first 100 moves were the finest anyone has ever played against the Master version of AlphaGo."

Yet it was not enough for Ke to beat this relative newcomer to the ancient board game, which has been around at least two millennia and enjoys enduring popularity in China, South Korea and Japan. DeepMind pursued mastery in the game notorious for its complexity, exceeding even that of chess. Lab founder and CEO Demis Hassabis says the number of possible positions in a game of Go outnumber the atoms in the universe.

And still Ke "pushed AlphaGo right to the limit," Hassabis tweeted.

"From the perspectives of human beings, he stretched a little bit and I was surprised," Ke said of his inhuman opponent at a news conference after Thursday's game.

"I also thought I was very close to winning the match in the middle of the game, but that might not have been what AlphaGo was thinking," he continued. "I was very excited, I could feel my heart thumping!"

The drama on the board did not unfold on broadcast, however.

Google has for years borne a fraught relationship with China, where the company's services have been unavailable since a 2010 dispute over the country's censorship policies. And that strain was on full display again this week, even as the game was not: Websites in China were not allowed to live-stream the match, according to apparent state instructions published by China Digital Times.

"This match may not be broadcast live in any form and without exception, including text commentary, photography, video streams, self-media accounts and so on," the instructions read. "No website (including sports and technology channels) or desktop or mobile apps may issue news alerts or push notifications about the course or result of the match."

At least two more matches await AlphaGo later this week: one more Saturday against Ke, for good measure, and another on Friday in which the program will take on a gang of five of the world's best players working together.

Continued here:

Google AI Clinches Series Against Humanity's Last, Best Hope To Win At Go - NPR

EVE Online has rolled out a challenging new computer-controlled enemy base that reacts to players strategies with artificial intelligence. A great reward lies at its center for players able to defeat it, although the first ones to do so werent able to collect.

On May 17th, CCP introduced a new type of NPC-pirate space station to its spacefaring MMO. The twist is that it can analyze player fleets that attempt to attack it, and use this information to spawn its own fleets in response that are specifically geared to defend against what it sees. If players bring overwhelming firepower in the form of massive capital ships, the AI will spawn its own capital vessels in response. If players bring a fleet of relatively smaller battleships, the AI will spawn hundreds of cruisers to swarm and neutralize the threat.

In addition to this adaptability, the AI is said to react and reposition its fleet in a manner similar to the best human fleet commanders. Blood Raider attack vessels will maneuver around the battlefield to cut off logistics and boost vessels, cutting off valuable support to the fleets.

If any players can best this new space station, they will be rewarded with expensive, rare blueprints for the games new Blood Raider capital class vessels. As only one such facility can ever be active on the live servers at a time, it becomes a hot ticket target, drawing attention from all over the galaxy.

The Red Alliance was the first group of players to attempt to destroy the space station, bringing to bear a fleet of powerful and expensive battleships. But the new AI proved too formidable, making short work of the fleet. News of the battle, along with the location of the hostile station, soon reached far and wide, drawing the attention of several major alliances who turned their focus on the new content.

Next up to bat, The Imperium launched their first assault against the Blood Raiders. Following in their allies footsteps, they fielded another massive fleet of Battleships to assault the complex and, after a short battle, began to suffer heavy losses and were forced to again follow their allies footsteps, this time in retreat.

Having seen two massively powerful and expensive fleets fail to take the station down, Imperium Fleet Commander Jay Amazingness decided the next attempt would employ a radical strategy: Punishers. The Punisher is one of the first ships that anyone playing EVE, even the Free to Play Alpha Clones, are able to fly, requiring little to no in-game experience or skill training. Jay was able to assemble a fleet of hundreds of Imperial pilots, each of them flying a hand-out frigate costing less than the ammunition expended by the previous battleship fleets.

Once arriving at the battlefield, the frigates set themselves into a wide orbit around the complex, each engaging their ships afterburners and accelerating them to the point that the guns on the Blood Raider reaction vessels could not track them well enough to ever land a hit. The player vessels orbited the space station for well over an hour, completely impervious to its firepower. Eventually, their low damage guns overwhelmed the shields on the massive structure, causing it to enter a 24 hour reinforcement window. The next day, more entry level frigates poured into the system from The Imperium, and once again, the Blood Raiders were powerless to stop them.

But things in EVE never go so smoothly. During the second battle, pilots from a cheekily-named rival group called Test Alliance Please Ignore were waiting to harass and delay the Imperial fleets efforts. Flying ships that were superior in every way to the Imperial Punishers, they had no trouble unleashing massive destruction, causing pain and frustration to the Imperial pilots. But the costs of the Punishers being destroyed were so low that the Imperium just continually replaced them, bringing more and more ships to the field without any real concern for their lifespan.

Test Alliances interference wasnt purely for the laughs: In the final moments of the space stations life, as the explosions began rocking the structure and it came apart under the constant fire from the Imperium fleet, hot shot Test pilot Hudders warped his ship directly into the wreck of the exploding citadel. In so doing, Hudders was able to steal the loot that dropped from the massive space station: a one-time use blueprint for making the Blood Raider Titan, the Molok.

It was the first ever of such a blueprint to drop, estimated at well over 100 billion ISK in value. Before he could escape the system, however, he was shot down by the enraged and somewhat embarrassed Imperium fleet. The prized blueprint was destroyed along with his ship.

Still, Hudders became an instant hero inside of the Test Allianceand to a lesser extent the rest of the EVE communityfor denying such a valuable prize to The Imperium. However, not all was lost from The Imperiums point of viewthey had outsmarted the new mechanic introduced into the game in under a week, succeeding where others had failed, and began readying their forces for when the second of such complexes spawn.

In addition, they proved that the content could be enjoyed by literally anyone who played EVE, even if they had only been playing for a few hours, as all players had been welcome into its massive Punisher fleet. CCP seemed pleased to see that the new content had caused such a massive reaction from their players. In an in-universe newsreel video they released (above), CCP even made a special nod to the fleets composition.

The future of the Engineering Complexes is being hotly debated right now, with changes being proposed by players from all sides, hoping to bring further balance to the new system.

Lee Yancy (@Sarin_Blackfist)has been an avid gamer for as long as he can remember, but ever since discovering them, he has found himself almost completely absorbed in MMO style games. EVE Online and World of Warcraft dominate the majority of his time.

See the rest here:

Devious EVE Online AI Finally Defeated By Hundreds Of Puny Ships - Kotaku

In BriefAs automation continues to infiltrate more and more aspects oflife, some are considering the potential of an AI taking over thepresidency. With a higher capacity for unbiased, constructiveproblem solving, an AI president could potentially prove to be abetter leader than our human ones. Robot President

The president. A highly coveted and highly controversial role, one traditionally held by humans. The use of traditionally is a recently added modifier, one necessarywhen discussing a Wired articlesuggesting that the role of president might one day be filled by an artificial intelligence (AI).

Others have pushed the idea of an AI president before, with one group even fighting for Watson, IBMs AI, to run for theposition in 2016. This massive, strange, sci-fi political makeover would completely upend the idea of traditional political discourse, so why are so many consideringit?

One major reason is level-headedness. Humans are vulnerable to emotion, obviously, but an AI can make decisions without prejudice, without anger, without resentment, without impulse, and without ego (a major concern with powerful positions). AI is, by nature, capable of fully considering all aspects of an issue before making up its mind. AI will not make quick decisions based off of a single fact, so in that respect, the machine has humanity beat.

Also, an AI president could not beinfluenced financially. An AI wouldnt have a financial stakein any businesses, so it wouldnt prioritize any over others. It would only calculate what would be best based on the most up-to-date existing facts. For example, an AI would not have any investmentin any energy source, so that wouldfreeit toconsider onlythe facts of the situation when determining action on climate change.

As with any radical idea, the election of an AI president seems impossible on the surface. Could we really ever advance to the point where it would be technologically feasible to entrust the presidency to an AI? Theoretically, yes. AI is advancing rapidly, so if a serious project to build the first presidential AI came to fruition, such a system could probably be built in the not-so-distant future.

The largest issue standing in the way of this becoming a future reality is whether or not we, as a people, will allow it. Would the idea ever be supported by a majority of people? Hard to imagine when you consider that the current political landscape appears more polarized than ever.

Asautomation creeps into more and more aspects of our lives, it is not too out-there to think that we might be willing to one day elect a president unswayed by human shortcomings and programmed to act according to a set of ideals. The president could be programmed to follow conservative or liberal agendas, or instead of electing an AI designed to reflect one political party or another, we could just vote on various issues that would then be considered by our unbiased AI leader.

As AI systems take on tasks traditionally held by humandoctors, lawyers, and even songwriters, considering an AI president isnt so far-fetched. AI can now even be used to create better AI, so it might be time for us to consider a future in which our political leaders are smarter, fairer,and, well, less human.

More:

Could an AI Ever Be Elected President? - Futurism

Forbes

Is Cybersecurity A Second Coming For AI? Forbes Artificial intelligence (AI) is all over the technology headlines lately. It seems to be the latest buzzword to take hold, yet the question remains: Will this be a quick fad, or are we actually seeing the second coming of AI? Most recently, there has ...

Read more:

Is Cybersecurity A Second Coming For AI? - Forbes

CDC

So far the results are promising. Using AI, Ribeiro and her colleagues were able to predict whether someone would attempt suicide within the next two years at about 80 percent accuracy, and within the next week at 92 percent accuracy. Their findings were recently reported

This high level of accuracy was possible because of machine learning, as researchers trained an algorithm by feeding it anonymous health records from 3,200 people who had attempted suicide. The algorithm learns patterns through examining combinations of factors that lead to suicide, from medication use to the number of ER visits over many years. Bizarre factors may pop up as related to suicide, such as acetaminophen use a year prior to an attempt, but that doesn't mean taking acetaminophen can be isolated as a risk factor for suicide.

"As humans, we want to understand what to look for," Ribeiro says. "But this is like asking what's the most important brush stroke in a painting."

With funding from the Department of Defense, Ribeiro aims to create a tool that can be used in clinics and emergency rooms to better find and help high-risk individuals.

Related:

But not everyone who commits suicide has an extensive medical record or has ever even walked into a hospital. So another research team is taking a similar AI-powered approach to suicide prediction by examining data from a more ubiquitous source: smartphones. The DARPA-funded company Cogito has developed a mobile app called Companion that automatically gathers data on someone's communication and movement patterns.

"We don't listen to the conversations, but look at things like how many calls you make, how many miscalls you have," says Skyler Place, a scientist at Cogito. "We are taking hundreds and thousands of data points, combining them in a way that humans can't."

The data is then used to create a risk score that is shown to the clinician. If that score changes over time, the clinician calls the individual to check in and see whether they need additional care. Cogito is currently working with the Department of Veterans Affairs at a suicide prevention center in Denver, Colorado to test the app with veterans who are at high risk for suicide. If the tests are successful, Place imagines more widespread use in a couple of years.

While an AI-powered tool may determine who is at risk of suicide, it cannot predict exactly when they might attempt it. Facebook is trying to solve this problem by

CDC

What happens after someone is found to be at risk is another challenge. Prevention strategies aren't always effective and it can be more difficult to protect high-risk individuals after they leave healthcare settings. To that end, researchers at the University of Vermont and Dartmouth have developed a system for both risk assessment and prevention.

First, the patients use

"They wouldn't have been caught if it were not for this tool," says Bill J. Hudenko, assistant professor of psychiatry at Dartmouth's Geisel School of Medicine.

The risk assessment tool is paired with

If a patient is classified as high risk, a nurse will help them get set up with the Proxi app to create a support system of friends and family, Hudenko adds. "Because we know what is most valuable in suicide prevention is to help them keep connections with others."

Follow NBC MACH on

See the rest here:

AI May Hold the Key to Stopping Suicide - NBCNews.com