Teacher Rishi Rawat has one student who is not human, but a machine.

Lessons take place at a lab inside the University of Southern California’s (USC) Clinical Science Center in Los Angeles, where Rawat teaches artificial intelligence, or AI.

To help the machine learn, Rawat feeds the computer samples of cancer cells.

“They’re like a computer brain, and you can put the data into them and they will learn the patterns and the pattern recognition that’s important to making decisions,” he explained.

AI may soon be a useful tool in health care and allow doctors to understand biology and diagnose disease in ways that were never humanly possible.

​Doctors not going away

“Machines are not going to take the place of doctors. Computers will not treat patients, but they will help make certain decisions and look for things that the human brain can’t recognize these patterns by itself,” said David Agus, USC’s professor of medicine and biomedical engineering, director at the Lawrence J. Ellison Institute for Transformative Medicine, and director at the university’s Center for Applied Molecular Medicine.

Rawat is part of a team of interdisciplinary scientists at USC who are researching how AI and machine learning can identify complex patterns in cells and more accurately identify specific types of breast cancer tumors.

Once a confirmed cancerous tumor is removed, doctors still have to treat the patient to reduce the risk of recurrence. The type of treatment depends on the type of cancer and whether the tumor is driven by estrogen. Currently, pathologists would take a thin piece of tissue, put it on a slide, and stain with color to better see the cells.

“What the pathologist has to do is to count what percentage of the cells are brown and what percentage are not,” said Dan Ruderman, a physicist who is also assistant professor of research medicine at USC.

The process could take days or even longer. Scientists say artificial intelligence can do something better than just count cells. Through machine learning, it can recognize complicated patterns on how the cells are arranged, with the hope, in the near future of making a quick and more reliable diagnosis that is free of human error.

“Are they disordered? Are they in a regular spacing? What’s going on exactly with the arrangement of the cells in the tissue,” described Ruderman of the types of patterns a machine can detect.

“We could do this instantaneously for almost no cost in the developing world,” Agus said.

​Computing power improves

Scientists say the time is ripe for the marriage between computer science and cancer research.

“All of a sudden, we have the computing power to really do it in real time. We have the ability of scanning a slide to high enough resolution so that the computer can see every little feature of the cancer. So it’s a convergence of technology. We couldn’t have done this, we didn’t have the computing power to do this several years ago,” Agus said.

Data is key to having a machine effectively do its job in medicine.

“Once you start to pool together tens and hundreds of thousands of patients and that data, you can actually [have] remarkable new insight, and so AI and machine learning is allowing that. It’s enabling us to go to the next level in medicine and really take that art to new heights,” Agus said.

Back at the lab, Rawat is not only feeding the computer more cell samples, he also designs and writes code to ensure that the algorithm has the ability to learn features unique to cancer cells.

The research now is on breast cancer, but doctors predict artificial intelligence will eventually make a difference in all forms of cancer and beyond.
…

Teacher Rishi Rawat has one student who is not human, but a machine.

Lessons take place at a lab inside the University of Southern California’s (USC) Clinical Science Center in Los Angeles, where Rawat teaches artificial intelligence, or AI.

To help the machine learn, Rawat feeds the computer samples of cancer cells.

“They’re like a computer brain, and you can put the data into them and they will learn the patterns and the pattern recognition that’s important to making decisions,” he explained.

AI may soon be a useful tool in health care and allow doctors to understand biology and diagnose disease in ways that were never humanly possible.

​Doctors not going away

“Machines are not going to take the place of doctors. Computers will not treat patients, but they will help make certain decisions and look for things that the human brain can’t recognize these patterns by itself,” said David Agus, USC’s professor of medicine and biomedical engineering, director at the Lawrence J. Ellison Institute for Transformative Medicine, and director at the university’s Center for Applied Molecular Medicine.

Rawat is part of a team of interdisciplinary scientists at USC who are researching how AI and machine learning can identify complex patterns in cells and more accurately identify specific types of breast cancer tumors.

Once a confirmed cancerous tumor is removed, doctors still have to treat the patient to reduce the risk of recurrence. The type of treatment depends on the type of cancer and whether the tumor is driven by estrogen. Currently, pathologists would take a thin piece of tissue, put it on a slide, and stain with color to better see the cells.

“What the pathologist has to do is to count what percentage of the cells are brown and what percentage are not,” said Dan Ruderman, a physicist who is also assistant professor of research medicine at USC.

The process could take days or even longer. Scientists say artificial intelligence can do something better than just count cells. Through machine learning, it can recognize complicated patterns on how the cells are arranged, with the hope, in the near future of making a quick and more reliable diagnosis that is free of human error.

“Are they disordered? Are they in a regular spacing? What’s going on exactly with the arrangement of the cells in the tissue,” described Ruderman of the types of patterns a machine can detect.

“We could do this instantaneously for almost no cost in the developing world,” Agus said.

​Computing power improves

Scientists say the time is ripe for the marriage between computer science and cancer research.

“All of a sudden, we have the computing power to really do it in real time. We have the ability of scanning a slide to high enough resolution so that the computer can see every little feature of the cancer. So it’s a convergence of technology. We couldn’t have done this, we didn’t have the computing power to do this several years ago,” Agus said.

Data is key to having a machine effectively do its job in medicine.

“Once you start to pool together tens and hundreds of thousands of patients and that data, you can actually [have] remarkable new insight, and so AI and machine learning is allowing that. It’s enabling us to go to the next level in medicine and really take that art to new heights,” Agus said.

Back at the lab, Rawat is not only feeding the computer more cell samples, he also designs and writes code to ensure that the algorithm has the ability to learn features unique to cancer cells.

The research now is on breast cancer, but doctors predict artificial intelligence will eventually make a difference in all forms of cancer and beyond.
…

With the help of artificial intelligence and machine learning, doctors may soon have new ways of diagnosing and treating patients in ways that were never humanly possible. Scientists at the University of Southern California in Los Angeles are developing a way of using machine learning to identify specific types of breast cancer tumors, and they say it’s just the beginning of what the computer can do. VOA’s Elizabeth Lee has the details from Los Angeles.
…

Microsoft’s big bet on cloud computing is paying off as the company has surpassed Apple as the world’s most valuable publicly traded company. 

The software maker’s prospects looked bleak just a few years ago, as licenses for the company’s Windows system fell with a sharp drop in sales of personal computers. 

But under CEO Satya Nadella, Microsoft has found stability by focusing on software and services over the internet, or the cloud, with long-term business contracts.  

That 1990s personal-computing powerhouse is now having a renaissance moment, as it eclipses Facebook, Google, Amazon and the other tech darlings of the late decade. 

Apple had been the world’s most prosperous firm since claiming the top spot from Exxon Mobil earlier this decade. Microsoft surpassed Apple briefly a few times this week, but didn’t close on top until Friday, with a market value of $851 billion to Apple’s $847 billion. Microsoft hadn’t been at the top since the height of the dot-com boom in 2000.  

Apple’s plunge

Microsoft became a contender again in large part because Apple’s stock fell nearly 20 percent in November, while Microsoft hasn’t done any worse than the rest of the stock market. But the fact that it hasn’t done poorly reflects its steady focus on business customers in recent years. 

Microsoft lost its luster as people were shunning PCs in favor of smartphones. In 2013, PC sales plunged 10 percent to about 315 million, the worst year-to-year drop ever, according to research firms Gartner and IDC. It didn’t help that Microsoft’s effort to make PCs more like phones, Windows 8, was widely panned.  

But a turnaround began when the Redmond, Wash., company promoted Nadella as CEO in 2014. He succeeded Microsoft’s longtime CEO, Steve Ballmer, who initially scoffed at the notion that people would be willing to pay $500 or more for Apple’s iPhones.  

That bet paid off. Windows is now a dwindling fraction of Microsoft’s business. While the company still runs consumer-focused businesses such as Bing search and Xbox gaming, it has prioritized business-oriented services such as its Office line of email and other workplace software, as well as newer additions such as LinkedIn and Skype. But its biggest growth has happened in the cloud, particularly the cloud platform it calls Azure. Cloud computing now accounts for more than a quarter of Microsoft’s revenue, and Microsoft rivals Amazon as a leading provider of such services. 

Wedbush analyst Dan Ives said Azure is still in its early days, meaning there’s plenty of room for growth, especially considering the company’s large customer base for Office and other products. 

“While the tech carnage seen over the last month has been brutal, shares of [Microsoft] continue to hold up like the Rock of Gibraltar,” he said. 

Being less reliant on consumer demand helped shield Microsoft from holiday season turbulence and U.S.-China trade war jitters affecting Apple and other tech companies. 

President Donald Trump amplified those tariff concerns when he told The Wall Street Journal in a story published late Monday that new tariffs could affect iPhones and laptops imported from China. 

The iPhone maker had already seen its stock fall after reporting a mixed bag of quarterly results in early November amid fears about how the technology industry will fare in the face of such threats as rising interest rates, increased government regulation and Trump’s escalating trade war with China. 

Reporting change

Apple also spooked investors with an unexpected decision to stop disclosing how many iPhones it sells each quarter. That move has been widely interpreted as a sign that Apple foresees further declines in iPhone sales and is trying to mask that.  

While smartphones caused the downturn in personal computers years ago, sales of smartphones themselves have now stalled. That’s partly because with fewer innovations from previous models, more people choose to hold on to the devices for longer periods before upgrading. 

Daniel Morgan, senior portfolio manager for Synovus Trust, said Microsoft is outperforming its tech rivals in part because of what it’s not. It doesn’t face as much regulatory scrutiny as advertising-hungry Google and Facebook, which have attracted controversy over their data-harvesting practices. Unlike Netflix, it’s not on a hunt for a diminishing number of international subscribers. And while Amazon also has a strong cloud business, it’s still more dependent on online retail.  
…

U.S. Deputy Attorney General Rod Rosenstein called on social media companies and technology firms Thursday to work with law enforcement to protect the public from cybercriminals.  

Speaking at a symposium on online crime, Rosenstein said that “social media platforms provide unprecedented opportunities for the free exchange of ideas. But many users do not understand that the platforms allow malicious actors, including foreign government agents, to deceive them by launching vast influence operations.” 

He said it was up to the companies to “place security on the same footing as novelty and convenience, and design technology accordingly.”  

He warned that if the technology sector failed to do so, government would have to step in.  

“I think the companies now do understand if they do not take it upon themselves to self-regulate — which is essentially the theme of my talk today — they will face the potential of government regulation,” he said. 

Extortion scheme

Rosenstein’s remarks came a day after the Justice Department charged two Iranian hackers in connection with a multimillion-dollar cybercrime and extortion scheme that targeted government agencies, cities and businesses. 

Rosenstein said many tech companies are willing to work with law enforcement and to prevent the use of their platforms to spread disinformation. 

But he said that “some technology experts castigate colleagues who engage with law enforcement to address encryption and similar challenges. Just because people are quick to criticize you does not mean that you are doing the wrong thing.” 

U.S. law enforcement officials have long been pushing tech companies to make it easier for them to access information on private devices such as cellphones and social media accounts. But most firms have resisted, citing privacy of the users.  

Rosenstein said data encryption practices were a “significant detriment to public safety.”  

“Improvements in the ability to investigate crime and hold perpetrators accountable must match the pace at which technology is making crimes easier to commit and more destructive,” Rosenstein said. 
…

Facebook is cautiously expanding a feature that shows people local news and information, including missing-person alerts, road closures, crime reports and school announcements.

Called “Today In,” the service shows people information from their towns and cities from such sources as news outlets, government entities and community groups. Facebook launched the service in January with six cities and expanded that to 25, then more. On Wednesday, “Today In” is expanding to 400 cities in the U.S. — and a few others in Australia.

The move comes as Facebook tries to shake off its reputation as a hotbed for misinformation and elections-meddling and rather a place for communities and people to come together and stay informed.

Here are some things to know about this effort, and why it matters:

The big picture

It’s something users have asked for, the company says. Think of it as an evolution of a “trending” feature the company dropped earlier this year. That feature, which showed news articles that were popular among users, but was rife with such problems as fake news and accusations of bias.

Anthea Watson Strong, product manager for local news and community information, said her team learned from the problems with that feature.

“We feel deeply the mistakes of our foremothers and forefathers,” she said.

This time around, Facebook employees went to some of the cities they were launching in and met with users. They tried to predict problems by doing “pre-mortem” assessments, she said. That is, instead of a “post-mortem” where engineers dissect what went wrong after the fact, they tried to anticipate how people might misuse a feature — for financial gain, for example.

Facebook isn’t saying how long it has been taking this “pre-mortem” approach, though the practice isn’t unique to the company. Nonetheless, it’s a significant step given that many of Facebook’s current problems stem from its failure to foresee how bad actors might co-opt the service.

Facebook also hopes the feature’s slow rollout will prevent problems.

How it works

To find out if “Today In” is available in your city or town, tap the “menu” icon with the three horizontal lines. Then scroll down until you see it. If you want, you can choose to see the local updates directly in your news feed.

For now, the company is offering this only in small and mid-sized cities such as Conroe, Texas, Morgantown, West Virginia, and Santa Fe, New Mexico. Large cities such as New York or Los Angeles have added challenges, such as an abundance of news and information, and may need to be broken up into smaller neighborhoods.

The posts in “Today In” are curated by artificial intelligence; there is no human involvement. The service aggregates posts from the Facebook pages for news organizations, government agencies and community groups like dog shelters. For this reason, a kid couldn’t declare a snow day, because “Today In” relies on the school’s official page. Discussion posts from local Facebook groups may also be included.

For now, the information is tailored only by geography, but this might change. A person with no kids, for example, might not want to see updates from schools.

Safeguards?

Facebook uses software filters to weed out objectionable content, just as it does on people’s regular news feed. But the filters are turned up for “Today In.” If a good friend posts something a bit objectionable, you are still likely to see it because Facebook takes your friendship into account. But “Today In” posts aren’t coming from your friends, so Facebook is more likely to keep it out.

 
…

Porsche says its future is in electric cars but for now it is rolling out a more powerful version of its internal combustion mainstay, the sleek 911 sports car.

Stuttgart-based Porsche, part of Volkswagen, is to show off the eighth version of its brand-defining model at the Los Angeles Auto Show.

The new 911 doesn’t look much different than earlier editions of the car. The new one has bigger wheel housings and a slightly wider body but the same long hood, sloping roof and prominent headlights that have marked successive versions since 1963.

The company said in a news release Wednesday that the new 911 Carrera S and 4S have flat six-cylinder turbocharged engines putting out 443 horsepower, 23 horsepower more than the predecessor. The Carrera S has a top speed of 191 mph and accelerates from zero to 60 mph (96.5 kph) in 3.5 seconds.

The rear-drive 2020 Carrera S has a base price of $113,200 and the 4S all-wheel drive version starts at $120,600, not including a $1,050 delivery fee. They can be ordered now and will reach dealers in summer 2019.

Porsche boss Oliver Blume says that the 911 remains “the core of our brand, we are making it even more emotional.”

Blume says nonetheless by 2025 about half of all new Porsche cars and SUVs will have electric motors, whether they are all-electric or hybrids combining batteries with internal combustion engines.

He was quoted by the Welt am Sonntag newspaper as saying that the company would be ready for a world in which some cities and countries are talking about banning internal combustion cars in coming decades. “It’s clear, the future belongs to electric mobility,” he said.

The company is developing an all-electric sports car, the Taycan, that would compete with sports car offerings by Tesla, BMW and others.

 
…

In the first case of its kind, the U.S. Justice Department announced charges Wednesday against two alleged Iranian cybercriminals who used malware to infect the computer networks of U.S. municipalities, hospitals and other organizations in a scheme to extort millions of dollars from the victims.

Faramarz Shahi Savandi, 34, and Mohammad Mehdi Shah Mansouri, 27, are accused of creating and deploying a sophisticated malware known as SamSam Ransomware to forcibly encrypt data on the computer networks of more than 200 organizations and other victims in the United States and Canada.   Savandi and Mansouri would then demand a ransom payment in the form of the virtual currency known as bitcoin in exchange for decryption keys for the encrypted data.  

In all, the two allegedly received more than $6 million in extortion payments.  Officials did not name the victims that made the payments  Other victims that refused to pay ransom suffered more than $30 million in lost data.  

The victims included state agencies, city governments and hospitals, including the City of Atlanta, the City of Newark, the Port of San Diego, the Colorado Department of Transportation, the University of Calgary in Calgary, Canada, and six U.S. public health related entities, according to the 26-page indictment. 

Deputy Attorney General Rod Rosenstein announced the six-count indictment against Saandi and Mansouri.  The two residents of Tehran remain at large and have been placed on the FBI’s wanted list.  Officials said the pair have no known ties to the Iranian government.  

The hacking and extortion scheme lasted nearly three years starting in January 2016.  The investigation started when a victim of the scheme came forward, officials aid.

“Every sector of our economy is a target of malicious cyber activity,” Rosenstein said.  “But the events described in this indictment highlight the urgent need for municipalities, public utilities, health care institutions, unviersities other public organizations to enhance their cyber security.”

Assistant Attorney General Brian A. Benczkowski said the indictment was the first ever “against criminal actors for deploying a for-profit- ransomware, hacking, and extortion scheme.   

Benczkowsi  said the Iranian hackers carefully targeted their victims.  In one instance, a few days before attacking the network of Kansas Heart Hospital, “the defendants conducted online searches concerning the hospital and accessed its website,” he said.

In recent months, U.S. prosecutors have charged a number of Iranian hackers, some with ties to the Iranian government, with cybercrimes.  In March, prosecutors charged nine Iranian hackers with penetrating the computer networks of hundreds of American and foreign universities and other institutions.

Ransomware has become a favorite tool of cybercriminals in recent years.  According to  a report by the cybersecurity firm Bitdefender, ransomware payments were expected to hit a record $2 billion in 2017.

 
…

Residents in Seoul discovered how fragile their telecommunications system was this past weekend when a fire disrupted service for millions. The government and the provider vowed to implement changes to avoid a repeat of the event, but the system failure demonstrated a need for greater redundancy and preparation for future natural and technological disasters.

The fire affected customers of KT, the nation’s second largest telecommunications company. They found themselves unable to make calls, access the internet, complete ATM or credit card transactions, and watch television. Local media also reported an elderly woman died when she fell ill and her husband wasn’t able to reach emergency services during the service outage.

Lee Manjong, chairman of the Korean Association for Terrorism Studies and professor of the department of Law & Police at Howon University, told VOA that while it is nearly impossible to prevent widespread system outages, certain steps can be taken to avoid catastrophic failures.

“It is necessary to split the public safety net (fire, medical, and police emergency services) and make system backups (redundancies) compulsory,” he said.

Following the blaze, South Korea’s minister of Science and ICT (Information, Communications, and Technology), You Young-min, spoke to the CEOs of South Korea’s three major communication companies (SK Telecom, KT, and LG U+) to discuss their backup plans.

You said the companies “need to swiftly change their contract clauses on compensation issues and also need to come up with plans that would reroute traffic if such accidents, which shouldn’t happen again, happen.”

When asked for specifics on what steps the government planned on taking to prevent a similar event in the future, the ministry declined to offer specifics, stating that responsible parties would prepare fire prevention measures this year and set up a task force to implement recommendations.

Local broadcaster MBC also reported that telecommunication companies and the government held a 20-minute virtual natural disaster drill in May to simulate a system outage, but the simulation proved to be ineffective in real-world situations.

Interconnected services

The Seoul fire and resulting system outage demonstrated how interconnected services are in the 21st century.

“If a network is down, then it affects other networks such as finance, power, energy, and railway,” said Lee.

He said there are multiple ways the electronic infrastructure can be paralyzed. This includes physical damage, natural disasters, and cyber attacks. However, Lee notes disruptions caused by cyber incursions are more effective.

“Cyber attacks are more efficient as they can take place without access to the physical location of the target,” he said.

According to Lee, this is because the government is able to secure physical sites, so cyber-warriors choose “soft targets” connected through the Internet.

A distributed denial-of-service (DDos) attack could be launched from the Internet and attack telecommunication networks. This type of attack floods a computer network with incoming data packets and overwhelms the system, effectively shutting it down. Lee said such attacks on telecom systems could wreak havoc and paralyze communication.

He cautioned that a successful cyber attack on South Korea’s technological infrastructure could yield “unimaginable” damage because of the country’s reliance on networked services.

Fire and recovery

Saturday’s fire struck an underground facility of KT, destroying telephone lines and fiber optic cables, taking about 10 hours to suppress. 

Seoul authorities rate facilities on a scale from A to D. Buildings rated A, B, or C must have adequate fire prevention systems installed, while those receiving a D rating do not.

KT’s Ahyeon facility, where the fire took place, was one of 27 D-rated facilities belonging to the company. As such, fire scene investigators found there were no fire detectors or sprinkler system installed at the Ahyeon facility and only a single fire extinguisher present.

South Korea’s other telecommunication carriers utilize over 800 similar facilities throughout the country, none of which are required to have fire detection equipment or sprinklers installed.

Lee said government regulations must be altered to bridge the gaps to ensure that such facilities are required to have redundant services elsewhere in the event of a natural disaster or cyber attack.

Estimates are the blaze resulted in about $7 million in property damage. KT has announced it would compensate affected customers by awarding them a free month of service for their inconvenience. KB security expects that amount to total about $27.5 million.

In a text message to customers, KT said it was “deeply sorry for the inconvenience. We will adopt preventive measures such as safety inspections… to avoid a recurrence.”

Seoul officials told VOA the cause of the fire remains unknown and the investigation to determine its source could last a month.

Lee Ju-hyun contributed to this report.
…

There are plenty of places to get a cup of coffee in San Francisco. But a new kind of café offers espressos and cappuccinos made by a robot. Michelle Quinn stopped by to see if a robot can make a good café latte.
…

Alphabet Inc’s Google in May introduced a slick feature for Gmail that automatically completes sentences for users as they type. Tap out “I love” and Gmail might propose “you” or “it.” But users are out of luck if the object of their affection is “him” or “her.”

Google’s technology will not suggest gender-based pronouns because the risk is too high that its “Smart Compose” technology might predict someone’s sex or gender identity incorrectly and offend users, product leaders revealed to Reuters in interviews.

Gmail product manager Paul Lambert said a company research scientist discovered the problem in January when he typed “I am meeting an investor next week,” and Smart Compose suggested a possible follow-up question: “Do you want to meet him?” instead of “her.”

Consumers have become accustomed to embarrassing gaffes from autocorrect on smartphones. But Google refused to take chances at a time when gender issues are reshaping politics and society, and critics are scrutinizing potential biases in artificial intelligence like never before.

“Not all ‘screw ups’ are equal,” Lambert said. Gender is a “a big, big thing” to get wrong.

Getting Smart Compose right could be good for business. Demonstrating that Google understands the nuances of AI better than competitors is part of the company’s strategy to build affinity for its brand and attract customers to its AI-powered cloud computing tools, advertising services and hardware.

Gmail has 1.5 billion users, and Lambert said Smart Compose assists on 11 percent of messages worldwide sent from Gmail.com, where the feature first launched.

Smart Compose is an example of what AI developers call natural language generation (NLG), in which computers learn to write sentences by studying patterns and relationships between words in literature, emails and web pages.

A system shown billions of human sentences becomes adept at completing common phrases but is limited by generalities. Men have long dominated fields such as finance and science, for example, so the technology would conclude from the data that an investor or engineer is “he” or “him.” The issue trips up nearly every major tech company.

Lambert said the Smart Compose team of about 15 engineers and designers tried several workarounds, but none proved bias-free or worthwhile. They decided the best solution was the strictest one: Limit coverage. The gendered pronoun ban affects fewer than 1 percent of cases where Smart Compose would propose something, Lambert said.

“The only reliable technique we have is to be conservative,” said Prabhakar Raghavan, who oversaw engineering of Gmail and other services until a recent promotion.

New policy

Google’s decision to play it safe on gender follows some high-profile embarrassments for the company’s predictive technologies.

The company apologized in 2015 when the image recognition feature of its photo service labeled a black couple as gorillas. In 2016, Google altered its search engine’s autocomplete function after it suggested the anti-Semitic query “are jews evil” when users sought information about Jews.

Google has banned expletives and racial slurs from its predictive technologies, as well as mentions of its business rivals or tragic events.

The company’s new policy banning gendered pronouns also affected the list of possible responses in Google’s Smart Reply. That service allow users to respond instantly to text messages and emails with short phrases such as “sounds good.”

Google uses tests developed by its AI ethics team to uncover new biases. A spam and abuse team pokes at systems, trying to find “juicy” gaffes by thinking as hackers or journalists might, Lambert said.

Workers outside the United States look for local cultural issues. Smart Compose will soon work in four other languages: Spanish, Portuguese, Italian and French.

“You need a lot of human oversight,” said engineering leader Raghavan, because “in each language, the net of inappropriateness has to cover something different.”

Wispread challenge

Google is not the only tech company wrestling with the gender-based pronoun problem. Agolo, a New York startup that has received investment from Thomson Reuters, uses AI to summarize business documents.

Its technology cannot reliably determine in some documents which pronoun goes with which name. So the summary pulls several sentences to give users more context, said Mohamed AlTantawy, Agolo’s chief technology officer.

He said longer copy is better than missing details. “The smallest mistakes will make people lose confidence,” AlTantawy said. “People want 100 percent correct.”

Yet, imperfections remain. Predictive keyboard tools developed by Google and Apple Inc propose the gendered “policeman” to complete “police” and “salesman” for “sales.”

Type the neutral Turkish phrase “one is a soldier” into Google Translate and it spits out “he’s a soldier” in English. So do translation tools from Alibaba and Microsoft Corp. Amazon.com Inc opts for “she” for the same phrase on its translation service for cloud computing customers.

AI experts have called on the companies to display a disclaimer and multiple possible translations.

Microsoft’s LinkedIn said it avoids gendered pronouns in its year-old predictive messaging tool, Smart Replies, to ward off potential blunders.

Alibaba and Amazon did not respond to requests to comment. Warnings and limitations like those in Smart Compose remain the most-used countermeasures in complex systems, said John Hegele, integration engineer at Durham, North Carolina-based Automated Insights Inc, which generates news articles from statistics.

“The end goal is a fully machine-generated system where it magically knows what to write,” Hegele said. “There’s been a ton of advances made but we’re not there yet.”
…

Facebook came under fire on Tuesday from lawmakers from several countries who accused the firm of undermining democratic institutions and lambasted chief executive Mark Zuckerberg for not answering questions on the matter.

Facebook is being investigated by lawmakers in Britain after consultancy Cambridge Analytica, which worked on Donald Trump’s presidential campaign, obtained the personal data of 87 million Facebook users from a researcher, drawing attention to the use of data analytics in politics.

Concerns over the social media giant’s practices, the role of political adverts and possible interference in the 2016 Brexit vote and U.S. elections are among the topics being investigated by British and European regulators.

While Facebook says it complies with EU data protection laws, a special hearing of lawmakers from several countries around the world in London criticized Zuckerberg for declining to appear himself to answer questions on the topic.

“We’ve never seen anything quite like Facebook, where, while we were playing on our phones and apps, our democratic institutions… seem to have been upended by frat-boy billionaires from California,” Canadian lawmaker Charlie Angus said.

“So Mr. Zuckerberg’s decision not to appear here at Westminster [Britain’s parliament] to me speaks volumes.”

Richard Allan, the vice president of policy solutions at Facebook who appeared in Zuckerberg’s stead, admitted Facebook had made mistakes but said it had accepted the need to comply with data rules.

“I’m not going to disagree with you that we’ve damaged public trust through some of the actions we’ve taken,” Allan told the hearing.

Facebook has faced a barrage of criticism from users and lawmakers after it said last year that Russian agents used its platform to spread disinformation before and after the 2016 U.S. presidential election, an accusation Moscow denies.

Allan repeatedly declined to give an example of a person or app banned from Facebook for misuse of data, aside from the GSR app which gathered data in the Cambridge Analytica scandal.

Legal documents reviewed by Reuters show how the investigation by British lawmakers has led them to seize documents relating to Facebook from app developer Six4Three, which is in a legal dispute with Facebook.

Damian Collins, chair of the culture committee which convened the hearing, said he would not release those documents on Tuesday as he was not in a position to do so, although he has said previously the committee has the legal power to.
…