Artificial intelligence: How to avoid racist algorithms

Bing image search results for babies - all white babiesImage copyright

Image caption

Only white babies appear in a search for “babies” on Microsoft search engine Bing…

There is growing concern that many of the algorithms that make decisions about our lives – from what we see on the internet to how likely we are to become victims or instigators of crime – are trained on data sets that do not include a diverse range of people.

The result can be that the decision-making becomes inherently biased, albeit accidentally.

Try searching online for an image of “hands” or “babies” using any of the big search engines and you are likely to find largely white results.

In 2015, graphic designer Johanna Burai created the World White Web project after searching for an image of human hands and finding exclusively white hands in the top image results on Google.

Her website offers “alternative” hand pictures that can be used by content creators online to redress the balance and thus be picked up by the search engine.

Google says its image search results are “a reflection of content from across the web, including the frequency with which types of images appear and the way they’re described online” and are not connected to its “values”.

Ms Burai, who no longer maintains her website, believes things have improved.

“I think it’s getting better… people see the problem,” she said.

“When I started the project people were shocked. Now there’s much more awareness.”

Image copyright

Image caption

…and white hands appear if you type “hands” into Google.

The Algorithmic Justice League (AJL) was launched by Joy Buolamwini, a postgraduate student at the Massachusetts Institute of Technology, in November 2016.

She was trying to use facial recognition software for a project but it could not process her face – Ms Buolamwini has dark skin.

“I found that wearing a white mask, because I have very dark skin, made it easier for the system to work,” she says.

“It was the reduction of a face to a model that a computer could more easily read.”

It was not the first time she had encountered the problem.

Five years earlier, she had had to ask a lighter-skinned room-mate to help her.

“I had mixed feelings. I was frustrated because this was a problem I’d seen five years earlier was still persisting,” she said.

“And I was amused that the white mask worked so well.”

Image copyright
Joy Buolamwini

Image caption

Joy Buolamwini found her computer system recognised the white mask, but not her face.

Ms Buolamwini describes the reaction to the AJL as “immense and intense”.

This ranges from teachers wanting to show her work to their students, and researchers wanting her to check their own algorithms for signs of bias, to people reporting their own experiences.

And there seem to be quite a few.

One researcher wanted to check that an algorithm being built to identify skin melanomas (skin cancer) would work on dark skin.

“I’m now starting to think, are we testing to make sure these systems work on older people who aren’t as well represented in the tech space?” Ms Buolamwini says.

“Are we also looking to make sure these systems work on people who might be overweight, because of some of the people who have reported it? It is definitely hitting a chord.”

Diverse data

Ms Buolamwini thinks the situation has arisen partly because of the well-documented lack of diversity within the tech industry itself.

Every year the tech giants release diversity reports and they make for grim reading.

You get the picture. But what has that got to do with algorithms?

“If you test your system on people who look like you and it works fine then you’re never going to know that there’s a problem,” Joy Buolamwini argues.

Biased beauty

Of the 44 winners of a beauty contest last year judged by algorithms, and based on some 6,000 uploaded selfies from 100 different countries, only one was non-white and a handful were Asian.

Alex Zhavoronkov, Beauty.AI’s chief science officer, told the Guardian the result was flawed because the data set used to train the AI (artifical intelligence) had not been diverse enough.

“If you have not that many people of colour within the data set, then you might actually have biased results,” he said at the time.

On a more serious note, AI software used in the US to predict which convicted criminals might reoffend, was found to be more likely to incorrectly identify black offenders as high risk and white offenders as low risk, according to a study by the website Propublica (the software firm disputed these findings).

Suresh Venkatasubramanian, an associate professor at the University of Utah school of computing, says creators of AI need to act now while the problem is still visible.

“The worst that can happen is that things will change and we won’t realise it,” he told the BBC.

“In other words the concern has been that the bias, or skew, in decision-making will shift from things we recognise as human prejudice to things we no longer recognise and therefore cannot detect – because we will take the decision-making for granted.”

Image copyright
Getty Images

Image caption

Are we accidentally programming prejudiced robots?

He is however optimistic about tech’s progress.

“To say all algorithms have racist manifestations doesn’t make sense to me,” he says.

“Not because it’s impossible but because that’s not how it’s actually working.

“In the last three to four years what’s picked up is the discussion around the problems and possible solutions,” he adds.

He offers a number of these:

  • creating better and more diverse data sets with which to train the algorithms (they learn by processing thousands of, for example, images)
  • sharing best practice among software vendors, and
  • building algorithms which explain their decision making so that any bias can be understood.

Ms Buolamwini says she is hopeful that the situation will improve if people are more aware of the potential problems.

“Any technology that we create is going to reflect both our aspirations and our limitations,” she says.

“If we are limited when it comes to being inclusive that’s going to be reflected in the robots we develop or the tech that’s incorporated within the robots.”

Article source:

‘World’s worst drawer’ stumps Google AutoDraw

Google has released an artificial intelligence-powered drawing program that turns rough scribbles into illustrations.

The BBC’s Rory Cellan-Jones is the self-proclaimed “world’s worst drawer”.

So, he was keen to see if AutoDraw could make any sense of his sketches.

Article source:

Aga app ‘could let hackers turn off oven’

Aga cookerImage copyright

Image caption

A security researcher found the issues when considering whether to upgrade to the latest Aga model

An app that lets Aga cooker owners remotely control their ovens could be hijacked by hackers, a cybersecurity researcher has claimed.

Ken Munro of Pen Test Partners was thinking of upgrading his Aga when he found vulnerabilities in the apps used to control the newest models.

It means ovens could be turned on or off, though not in a way that makes the cookers dangerous.

Aga has said it has contacted the third party that provided the system.

“If you were maliciously motivated, it wouldn’t be very difficult to switch off people’s Aga’s remotely,” Mr Munro told the BBC.

His investigation concerned the “iTotal Control” (TC) system, which Aga has marketed since 2012.

Among the security issues he says he found is the fact that SMS messages – which are used by the system to turn the oven on or off – are not authenticated by the cooker.

Nor is the Sim card set up to send the messages validated on registration.

Mr Munro also criticised the fact that user registration for the service allows passwords as short as five characters – security experts usually recommend using as many characters as possible, with a minimum of eight.

Email addresses are sent in plain text via the system, too, he explained – meaning personal data could be vulnerable to snoopers.

Image copyright

Image caption

the mobile and web app allows user registration with a very short, five character, password

He also said that attempts to contact Aga about the problems, including a tweet and emails on 3 April, fell on deaf ears.

When he did get through to someone and advised them to take the Total Control website down, he got a disappointing response.

“I asked to speak to relevant departments, they couldn’t put me through,” he said.

Third party provider

“Aga Rangemaster operates its Aga TC phone app via a third party service provider,” Aga said in a statement.

“Security and account registration also involves our [machine to machine] provider.

“We take such issues seriously and have raised them immediately with our service providers so that we can answer in detail the points raised.”

However, the firm did not comment on Mr Munro’s claims that it ignored his disclosure of the problems.

Image copyright
Ken Munro

Image caption

The Aga cookers are controlled via SMS messages sent via the remote control system

“It’s kind of unacceptable that some random person could just take control of your Aga,” said Professor Alan Woodward, a cybersecurity expert at the University of Surrey.

“Will hackers try it? Who knows, but it just shouldn’t be possible.”

He added that he was surprised there seemed to be a flat response from the firm when Mr Munro tried to raise the issues.

“If somebody calls up, ‘I found a problem with your system,’ they should look at it,” Prof Woodward told the BBC.

Article source:

Callisto Group hackers targeted Foreign Office data

Foreign Office websiteImage copyright

The UK’s Foreign Office was targeted by highly motivated and well-resourced hackers over several months in 2016.

The BBC understands the government has investigated the previously unreported attack that began in April last year.

The UK’s National Cyber Security Centre would not say whether data was stolen.

But a source told the BBC that the most sensitive Foreign Office information is not kept on the systems targeted by the hackers.

Research published on Thursday by cybersecurity firm F-Secure suggested the attack was a “spear-phishing” campaign, in which people were sent targeted emails in attempts to fool them into clicking a rogue link or handing over their username and password.

To do this, the attackers created a number of web addresses designed to resemble legitimate Foreign Office websites, including those used for accessing webmail.

F-Secure does not know whether the attack was successful.

The company says the domains were created by hackers that it calls the Callisto Group, which it says is still active.

Image copyright
Getty Images

Image caption

Callisto Group had attacked military personnel, government officials and journalists according to F-Secure

However the UK’s National Cyber Security Centre (NCSC) declined to say who was behind the attack on the Foreign Office.

In a statement, it said: “The first duty of government is to safeguard the nation and as the technical authority on cyber security, the NCSC is delivering ground breaking innovations to make the UK the toughest online target in the world.

“The government’s Active Cyber Defence programme is developing services to block, prevent and neutralise attacks before they reach inboxes,” it added.


F-Secure said the Callisto Group had, since 2015, attacked “military personnel, government officials, think tanks and journalists” mainly in Eastern Europe and the South Caucasus, as well as in the Ukraine and the UK.

It added that there was some evidence suggesting the Callisto Group had ties with a nation state.

The company did not say which country, but also observed that the “infrastructure” used by the group had links with “entities” in China, Russia and Ukraine.

The targeted emails that were sent out tried to fool targets into downloading malware which was first developed for law enforcement by the Italian software company Hacking Team.

Hacking Team’s surveillance tools were previously exposed in a cyberattack, first reported in 2015.

There is no suggestion that Hacking Team had any involvement in the attacks.

F-Secure said that the use of the software should remind governments that they “don’t have monopolies on these [surveillance] technologies”, and that once created the software can fall into the hands of hackers.

Nation-state links?

The BBC has not seen evidence conclusively identifying the origin of the attack.

A cybersecurity expert at another company, who wished to remain anonymous, found a link to information uncovered in the investigation of Russian efforts to influence the US election.

Two of the phishing domains used by the hackers were once linked to an IP address mentioned in a US government report into Grizzly Steppe.

Grizzly Steppe is the name given by the US government to efforts by “Russian civilian and military intelligence services to compromise and exploit networks and endpoints associated with the US election”.

However, the cybersecurity expert noted that this connection between the phishing domain and Grizzly Steppe may be a coincidence, as over 300 other domains – many of them not hacking-related – were linked to the same IP address.

F-Secure told the BBC that it did notice some similarity between the Callisto Group’s hacking and previous attacks that have been linked to Russia.

However, it said despite some similarities in the tactics, techniques, procedures and targets of the Callisto Group, and the Russia-linked group known as APT28, it believed the two were “operationally” separate.

It noted that the Callisto Group was also less “technically capable” than APT28.

Article source:

Imax reveals European cinema expansion plan

ImaxImage copyright
Getty Images

Image caption

Imax presents 2D and 3D movies on screens larger than the norm

Imax has struck a deal to more than double the number of its large screens in European cinemas run by the continent’s biggest operator.

AMC – which owns Odeon among other brands – plans to create 25 new Imax theatres over the next three years, about 10 of which will be in the UK.

Some will also be built in Germany, Italy, Spain and Scandinavia.

One expert said Imax had proved popular with those seeking a “premium experience”.

AMC and Imax expect to jointly spend about $25m (£20m) on the expansion plan and will subsequently share the revenues.

“One of the things that has really frustrated me over my 23 years [in charge] has been our inability to properly penetrate Europe,” Imax’s chief executive, Rich Gelfond, told the BBC.

“Europe has been challenging because of its infrastructure – the theatres are smaller, they are dated and don’t really fit the Imax specs, and the build-out [of new cinemas] has been slow.

“So, today is really a historic day for us.”

AMC is making the investment five months after taking over Odeon and a month after its acquisition of Nordic Cinemas. The company is itself controlled by China’s Dalian Wanda Group.

It is now engaged in a wider effort to upgrade its cinemas.

“One of the ways you get people not to watch a movie in their home or on their iPad is you offer them a 60ft [18.3m] screen,” said AMC’s chief executive Adam Aron.

“When consumers have so many choices… one of the smartest things we can do is put in more and more Imax or Imax-like experiences, because that is one of the long-term guarantors that people are going to still go out.”

The initiative will bring the total number of Imax cinemas run by all chains in Europe close to 250.

However, that figure will still lag behind the 800 in existence or in development in China and the 350 in the US.

Image copyright
Getty Images

Image caption

China’s cinema-building spree has helped Imax spread quickly across the country

“Imax has already proven its worth in many countries, and wouldn’t have built the brand it has, unless it had been able to attract the public,” commented Robert Mitchell from the trade magazine Variety.

“The question mark is whether it will succeed in countries like Germany, where cinema attendance has been dropping for the last couple of years and [where] Imax only has a few screens at present, so is less well known.”

Blockbuster films

Imax has developed a projection system that allows images to be shown on screens up to 100ft wide without becoming blurry.

It also offers filmmakers cameras that are able to capture higher resolution footage than normal.

Directors including Christopher Nolan and Michael Bay are fans, but the tech is normally restricted to blockbuster projects because of the additional costs it incurs.

Image copyright

Image caption

Imax cameras have been used on Dunkirk, Star Wars: The Force Awakens and Beauty and the Beast among other films

In addition, Imax offers software to convert footage shot with normal cinema cameras, and has developed a proprietary sound system for its theatres.

It faces competition from other large-screen formats including Dolby Cinema, RealD Luxe and an in-house effort from AMC.

Marvel exclusive

To help distinguish itself, Imax recently announced plans to start screening original content unavailable elsewhere.

Image copyright

Image caption

The first two episodes of The Inhumans have been filmed with Imax digital cameras

The first “exclusive” is a deal to show the first two episodes of a forthcoming Marvel TV show – The Inhumans – in September, several weeks before they come to television.

“Other companies like HBO went from Hollywood-content only to original programming,” Mr Gelfond explained.

“We’re following in those footsteps and thinking: why not bring a great original content experience at a slower time of the year?”

In addition, Imax recently launched a standalone virtual reality centre in Los Angeles and will shortly open different versions of the concept attached to cinemas in New York and London.

Members of the public pay to try out short experiences using StarVR headsets, which offer a wider field-of-view than rival tech.

The firm already offers content linked to the movies John Wick, Star Wars and The Walk, and has struck a deal with Warner Bros to create new experiences based on its DC superhero films.

Image copyright

Image caption

Imax’s range of VR content includes Star Wars: Trials on Tatooine, in which players wield lightsabers

Mr Gelfond said he hoped that this would attract Millennials – those born after mid-1990 – who had proved harder to attract to cinemas than older age groups.

But he acknowledged that it was early days for the tech.

“We are starting with about five to 10 VR centres that will all be open this year,” he said.

“We’ll look at the results. If they look good we’ll build it out at a much more significant way.

“And if they don’t we’ll have not lost much.”

Article source:

AI wins $290,000 in Chinese poker competition

Poker gameImage copyright
Getty Images

Image caption

A team of six human players were beaten by the artificial intelligence system

An artificial intelligence program has beaten a team of six poker players at a series of exhibition matches in China.

The AI system, called Lengpudashi, won a landslide victory and $290,000 (£230,000) in the five-day competition.

It is the second time this year that an AI program has beaten competitive poker players.

An earlier version of the program, known as Libratus, beat four of the world’s best poker pros during a 20-day game in January.

The AI systems were the work of Tuomas Sandholm, a computer science professor at Carnegie Mellon University in the US, and PhD student Noam Brown.

The prize money will go to Strategic Machine, a firm founded by the duo.

The human team up against Lengpudashi was led by Yue Du, an amateur poker player who won the World Series of Poker $5,000 buy-in, no-limit, Texas Hold’em category last year.

Mr Du’s “Team Dragon” consisted of engineers, computer scientists and investors who attempted to use game theory and their knowledge of machine intelligence to anticipate and counter Lengpudashi’s play.

‘Imperfect information’

Unlike chess and Go, in which all the playable pieces are visible on the board, poker is what computer scientists call an “imperfect information game”.

This means relying on complicated betting strategies and a player’s ability to bluff, or spot when opponents are bluffing.

“People think that bluffing is very human,” Mr Brown told Bloomberg, “It turns out that’s not true.”

“A computer can learn from experience that if it has a weak hand and it bluffs, it can make more money.”

Like the poker pro-beating Libratus AI program before it, Lengpudashi was run on a supercomputer at Pittsburgh Supercomputing Center.

Researchers commonly use matches like these to hone an AI program’s reasoning skills and strategic decision making.

Article source:

Microsoft patches serious Word bug ‘targeted by scammers’

Microsoft logoImage copyright
Getty Images

A bug in Word apparently targeted by scammers trying to steal banking logins will be patched, Microsoft has said.

The previously undetected, or “zero-day”, vulnerability had been reported over the weekend.

Then, on 10 April, cybersecurity firm Proofpoint announced it had discovered an email campaign targeting the bug that aimed to distributed Dridex malware.

Dridex is designed to infect a victim’s computer and snoop on banking logins.

In 2015, it was cited as the means by which cyber-attackers stole more than £20m from British bank accounts.

The flaw discovered in many versions of Microsoft Word for Windows could allow malicious software, including Dridex, to be installed, according to cybersecurity researchers.

Microsoft did not confirm whether Mac versions of Word were also affected.

A scam email campaign was found to be distributing Microsoft Word RTF [Rich Text Format] documents to recipients that contained Dridex.

‘Fully exploited’

“During our testing (for example on Office 2010) the vulnerable system was fully exploited,” wrote Proofpoint researchers in a blog.

“We plan to address this through an update on Tuesday April 11, and customers who have updates enabled will be protected automatically,” said a Microsoft spokesman.

“Meanwhile we encourage customers to practise safe computing habits online, including exercising caution before opening unknown files and not downloading content from untrusted sources to avoid this type of issue.”

Proofpoint also urged Microsoft Word users to install the security updates quickly.

“Because of the widespread effectiveness and rapid weaponisation of this exploit, it is critical that users and organisations apply the patch as soon as it becomes available,” the firm said.

Article source:

Dallas warning sirens ‘set off by hacker’

Warning sirensImage copyright

Image caption

Many US cities have siren systems for warning residents about hurricanes or tornadoes

A hacker has been blamed for setting off more than 150 warning sirens in the US city of Dallas over the weekend.

The sirens are usually used to warn of extreme weather events such as tornadoes.

“All 156 sirens in the city were activated last night – it does appear at this time that it was a hack,” a spokeswoman for the city told reporters.

The noise “woke up a lot of people”, she added.

The sirens were activated at 23:42 local time (04:42 GMT) on Friday and lasted for about 90 minutes.

Some posted footage online in which the sirens can clearly be heard.

Technicians for the Office of Emergency Management were eventually able to shut the warning system down and find what they said was evidence that the siren system had been hacked.

“We do believe that the hack came from the Dallas area,” a city statement said.

Last year, someone hacked into a number of traffic signs in Dallas and used them to publish jokes. There has been no suggestion that the same people were involved in the sirens incident.

Article source:

Google’s AI seeks further Go glory

GoImage copyright
Getty Images

Image caption

AlphaGo will challenge some of China’s top Go players later this month

Google has challenged China’s top Go player to a series of games against its artificial intelligence technology.

It said the software would play a best-of-three match against Ke Jie, among other games against humans in the eastern Chinese city of Wuzhen from 23-27 April.

Last year, the Google program recorded a 4-1 victory against one of South Korea’s top Go players.

One expert said that result had come as a surprise.

“A lot of AI researchers have been working on Go because it’s the most challenging board game we have,” said Calum Chace, author of Surviving AI.

“The conventional wisdom was that machines would ultimately triumph but it would take 10 years or so.

“The win was a big wake-up call for a lot of people, including many outside the AI community.”

Image copyright

Image caption

AlphaGo won four matches out of five against Lee Se-dol

Google’s AlphaGo software was developed by British computer company DeepMind, which was bought by the US search firm in 2014.

Its defeat of Lee Se-dol in March 2016 is seen as a landmark moment, similar to that of IBM’s Deep Blue AI beating Garry Kasparov at chess in 1997.

Several of the moves AlphaGo made defied conventional wisdom but ended up paying off.

However, many Go aficionados did not recognise Mr Lee as the world’s top player at the time of the contest.

So, the new competition against 19-year-old Mr Ke – who is the current number one according to a popular but unofficial player-ranking system – has the potential to bring additional prestige to Google.

“We’ve been hard at work improving AlphaGo to become even more creative, and since playing Lee Se-dol, the program has continued to learn through self-play training,” a spokeswoman for DeepMind told the BBC.

“We intend to publish more scientific papers in the future, which will include further details of AlphaGo’s progress.”

Google added that Mr Lee would also be invited, but was not sure if he would attend.

Image copyright

Image caption

Ke Jie – seen on the far right – met Google chief executive Sundar Pichai in Beijing last year

In addition to the games against Mr Ke, AlphaGo will also:

  • play games involving one Chinese pro facing off against another, each of whom will have an AlphaGo-powered virtual teammate
  • challenge a five-person team containing some of China’s top players, who will work together to try to beat the AI

Over the past year, DeepMind’s technology has also been used to find ways to reduce energy bills at Google’s data centres as well as to try to improve care in British hospitals.

Media captionGoogle’s DeepMind division beat Europe’s Go champion in October 2016

A fresh wave of positive publicity could help Google find further uses for its tech.

“If it loses this match, a lot of people will be delighted to claim that Google and DeepMind has overpromised and that this is the kind of hype we always get with AI,” commented Mr Chace.

“But I wouldn’t have thought Google is taking a huge risk.

“[In recent months] it has been playing a lot of very, very good AlphaGo players online without disclosing it was an AI playing, and has won all of the games.

“Unless this Ke Jie is some magnitude better than Lee Se-dol, I would think they are confident of winning.”

What is Go?

Media captionA brief guide to Go

Go is thought to date back to several thousand years ago in China.

Using black-and-white stones on a grid, players gain the upper hand by surrounding their opponent’s pieces with their own.

The rules are simpler than those of chess, but a player typically has a choice of 200 moves, compared with about 20 in chess – there are more possible positions in Go than atoms in the universe, according to DeepMind’s team.

That means a computer cannot win simply via brute force – searching through the consequences of millions of moves in seconds.

It can be very difficult to determine who is winning, and many of the top human players rely on instinct.

To prepare for its victory over Lee Se-dol, DeepMind trained its software on 30 million expert moves and then set the machine to play against itself millions of times to get a sense of what strategies worked.

The result was that some of the innovative moves AlphaGo made in its landmark match were described as being “beautiful” and highly unusual by observers.

Article source:

Russian arrested in Spain ‘over US election hacking’

Donald Trump on the campaignImage copyright
Getty Images

Image caption

Mr Levashov’s arrest relates to alleged “hacking” of the US election last year

Spanish police have arrested a Russian programmer for alleged involvement in “hacking” the US election, Spanish press reports have said.

Pyotr Levashov, arrested on 7 April in Barcelona, has now been remanded in custody.

A “legal source” also told the AFP news agency that Mr Levashov was the subject of an extradition request by the US.

The request is due to be examined by Spain’s national criminal court, the agency added.

El Confidencial, a Spanish news website, has said that Mr Levashov’s arrest warrant was issued by US authorities over suspected “hacking” that helped Donald Trump’s campaign.

Several cybersecurity experts, including Brian Krebs, have also linked Mr Levashov to a Russian spam kingpin, who uses the alias Peter Severa.

Article source: