Anyone who’s spent much time on the web will know what TL;DR stands for⁠—but, for everyone else, it’s an acronym for “Too Long, Didn’t Read”.

It’s an understandable sentiment we’ve all felt at some point. People lead busy lives. Some outlets now even specialise in short, at-a-glance news.

The problem is, it’s hard to get the full picture of a story in just a brief snippet.

In a world where fake news can be posted and spread like wildfire across social networks – almost completely unchecked – it feels even more dangerous to normalise “news” being delivered in short-form without full context.

There are two sides to most stories, and it’s hard to see how both can be summarised properly.

News folks are not going to love this. Some product manager announced internally a tool that FB is developing called "TL;DR". Basically, it will use AI to summarize long form articles and spit out bullet points so people don't have to read the full piece.

What could go wrong!

— editor Mac 🙃 (@RMac18) December 15, 2020

However, the argument also goes the other way. When articles are too long, people have a natural habit of skim-reading them. Skimming in this way often means people then believe they’re fully informed on a topic… when we know that’s often not the case.

TL;DR needs to strike a healthy balance between summarising the news but not so much that people don’t get enough of the story. Otherwise, it could increase existing societal problems with misinformation, fake news, and lack of media trust.

According to BuzzFeed, Facebook showed off TL;DR during an internal meeting this week. 

Facebook appears to be planning to add an AI-powered assistant to TL;DR which can answer questions about the article. The assistant could help to clear up anything the reader is uncertain about, but it’s also going to have to prove it doesn’t suffer from any biases which arguably all current algorithms suffer from to some extent.

The AI is also going to have to be very careful in not taking things like quotes out-of-context and end up further automating the spread of misinformation.

There’s also going to be a debate over what sources Facebook should use. Should Facebook stick only to the “mainstream media” which many believe follow the agendas of certain powerful moguls? Or serve news from smaller outlets without much historic credibility? The answer probably lies somewhere in the middle, but it’s going to be difficult to get right.

Facebook continues to be a major source of misinformation – in large part driven by algorithms promoting such content – and it’s had little success so far in any news-related efforts. I think most people will be expecting this to be another disaster waiting to happen.

(Image Credit: Mark Zuckerberg by Alessio Jacona under CC BY-SA 2.0 license)

The post Facebook is developing a news-summarising AI called TL;DR appeared first on AI News.

Cambridge Analytica was accused in March 2018 of using AI tools and big data to influence the results of the Brexit referendum and the US presidential election. Most objective observers probably felt the case was overblown, but it’s taken until now to be confirmed.

“From my review of the materials recovered by the investigation I have found no further evidence to change my earlier view that CA [Cambridge Analytica] was not involved in the EU referendum campaign in the UK,” wrote Information Commissioner Elizabeth Denham.

Cambridge Analytica did obtain a ton of user data—but through predominantly commercial means, and of mostly US voters. Such data is available to, and has also been purchased by, other electoral campaigns for targeted advertising purposes (the Remain campaigns in the UK actually outspent their Leave counterparts by £6 million.)

“CA were purchasing significant volumes of commercially available personal data (at one estimate over 130 billion data points), in the main about millions of US voters, to combine it with the Facebook derived insight information they had obtained from an academic at Cambridge University, Dr Aleksandr Kogan, and elsewhere,” wrote Denham.

The only real scandal was Facebook’s poor protection of users which allowed third-party apps to scrape their data—for which it was fined £500,000 by the UK’s data protection watchdog.

It seems the claims Cambridge Analytica used powerful AI tools were also rather overblown, with the information commissioner saying all they found were models “built from ‘off the shelf’ analytical tools”.

The information commissioner even found evidence that Cambridge Analytica’s own staff “were concerned about some of the public statements the leadership of the company were making about their impact and influence.”

Cambridge Analytica appears to have been a victim of those unable to accept democratic results combined with its own boasting of capabilities that weren’t actually that impressive.

You can read the full report here (PDF)

(Photo by Christian Lue on Unsplash)

The post Information Commissioner clears Cambridge Analytica of influencing Brexit appeared first on AI News.

Facebook has been its usual blessing and a curse during the pandemic. On the one hand, it’s helped people to stay connected and organise community efforts. On the other, it’s allowed dangerous misinformation to spread like wildfire that’s led to the increase in anti-vaccine and anti-mask movements.

The social media giant is hoping that AI can help to swing the balance more towards Facebook having an overall benefit within our communities.

If a person has posted asking for help because they’re unable to leave the house, Facebook’s AI may automatically match that person with someone local who has recently said they’re willing to get things like groceries or prescriptions for people.

In a blog post, Facebook explains how it built its matching algorithm:

We built and deployed this matching algorithm using XLM-R, our open-source, cross-lingual understanding model that extends our work on XLM and RoBERTa, to produce a relevance score that ranks how closely a request for help matches the current offers for help in that community.

The system then integrates the posts’ ranking score into a set of models trained on PyText, our open-source framework for natural language processing.

It’s a great idea which could go a long way to making a real positive impact on people in difficult times. Hopefully, we’ll see more of such efforts from Facebook to improve our communities.

(Photo by Bohdan Pyryn on Unsplash)

The post Facebook uses AI to help people support each other appeared first on AI News.

Voter manipulation is a concern for any functioning democracy and deepfakes provide a whole new challenge for social media platforms.

A fake video of Nancy Pelosi, Speaker of the United States House of Representatives, went viral last year after purportedly showing her slurring her words as if she was intoxicated. The clip shows how even a relatively unsophisticated video (it wasn’t an actual deepfake) could be used to cause reputational damage and swing votes.

Facebook refused to remove the video of Nancy Pelosi and instead said it would display an article from a third-party fact-checking website highlighting that it’s been edited and take measures to limit its reach. The approach, of course, was heavily criticised.

The new rules from Facebook claim that deepfake videos that are designed to be misleading will be banned. The problem with the rules is they don’t cover videos altered for parody or those edited “solely to omit or change the order of words,” which will not sound encouraging to those wanting a firm stance against manipulation.

In the age of “fake news,” many people have become aware not to necessarily believe what they read. Likewise, an increasing number of people also know how easily images are manipulated. Deepfake videos pose such a concern because the wider public are not yet aware enough of their existence or how to spot them.

A report from the NYU Stern Center for Business and Human Rights last September, covered by our sister publication MarketingTech, highlighted the various ways disinformation could be used ahead of this year’s presidential elections.

One of the eight predictions is that deepfake videos will be used “to portray candidates saying and doing things they never said or did”. Another prediction is that Iran and China may join Russia as sources of disinformation, the former perhaps now being even more likely given recent escalations between the US and Iran and the desire for non-military retaliation.

Legislation is being introduced to criminalise the production of deepfakes without disclosing that they’ve been modified, but the best approach is to limit them from being widely shared in the first place.

“A better approach, and one that avoids the danger of overreaching government censorship, would be for the social media platforms to improve their AI-screening technology, enhance human review, and remove deepfakes before they can do much damage,” the report suggests.

The month after Facebook refused to remove the edited video of Pelosi, a deepfake created by Israeli startup Canny AI aimed to raise awareness of the issue by making it appear like Facebook CEO Mark Zuckerberg said: “Imagine this for a second: One man, with total control of billions of people’s stolen data, all their secrets, their lives, their futures.”

Canny AI’s deepfake was designed to be clearly fake but it shows how easy it’s becoming to manipulate people’s views. In a tense world, it’s not hard to imagine what devastation could be caused simply by releasing a deepfake of a political leader declaring war or planning to launch a missile.

Interested in hearing industry leaders discuss subjects like this? , , , AI &

The post Facebook pledges crackdown on deepfakes ahead of the US presidential election appeared first on AI News.

Live-streaming poses a very modern but serious problem. As evident with the white supremacist attack in Christchurch in March, live-streams of terrorism can quickly go viral and taking down videos manually is often a slow process.

Automation can step-in to help identify and remove content to prevent terrorists from achieving the attention they seek. In order to teach its AI how to spot shootings, Facebook is teaming up with police in the UK.

Stephanie McCourt, Facebook’s Law Enforcement Outreach Lead (UK), said:

“Facebook’s work tackling threats from terrorism and extremism never stops. We invest heavily in people and technology to keep people safe on our platforms. But we can’t do it alone.

This partnership with the Met Police will help train our AI systems with the volume of data needed to identify these incidents. And we will remain committed to improving our detection abilities and keeping harmful content off Facebook.”

The police are providing first-person perspective videos from officers’ firearms training to Facebook. The clips will help the social media giant train its algorithms to detect shootings on live-streams.

Not only will early detection enable Facebook to more quickly remove terrorist videos, but it will also help to alert the police faster. The quicker a response can be deployed, the more lives can be potentially saved.

The police became involved with the project through the Counter-Terrorism Internet Referral Unit’s (CTIRU) long-standing relationship with Facebook.

Neil Basu, the UK’s top-ranking counter-terrorism officer, commented:

“As a result of the unit’s relationship with Facebook, coupled with the world-renowned expertise of the Met Police Firearms Command, the Met has been invited to take part in this innovative project.

The technology Facebook is seeking to create could help identify firearms attacks in their early stages and potentially assist police across the world in their response to such incidents.

Technology that automatically stops live streaming of attacks once identified, would also significantly help prevent the glorification of such acts and the promotion of the toxic ideologies that drive them.”

Officers from the world-renowned Firearms Command regularly train to respond to a wide variety of scenarios. These scenarios can include hostage situations carried out on land, water, public transport, and more.

Such a variety of footage will help Facebook train its AI to also recognise a large number of active shooter situations.

The post Facebook will teach its AI to spot shootings using videos of UK police firearms training appeared first on AI News.

The consortium includes Google DeepMind, Facebook AI, New York University, and the University of Washington. Each of the consortium’s members believe a more comprehensive benchmark is needed for NLP than current solutions.

The result is a benchmarking platform called SuperGLUE which replaces an older platform called GLUE with a “much harder benchmark with comprehensive human baselines,” according to Facebook AI. 

SuperGLUE helps to put NLP abilities to the test where previous benchmarks were beginning to pose too simple for the latest systems.

“Within one year of release, several NLP models have already surpassed human baseline performance on the GLUE benchmark. Current models have advanced a surprisingly effective recipe that combines language model pretraining on huge text data sets with simple multitask and transfer learning techniques,” Facebook said.

In 2018, Google released BERT (Bidirectional Encoder Representations from Transformers) which Facebook calls one of the biggest breakthroughs in NLP. Facebook took Google’s open-source work and identified changes to improve its effectiveness which led to RoBERTa (Robustly Optimized BERT Pretraining Approach).

RoBERTa basically “smashed it,” as the kids would say, in commonly-used benchmarks:

“Within one year of release, several NLP models (including RoBERTa) have already surpassed human baseline performance on the GLUE benchmark. Current models have advanced a surprisingly effective recipe that combines language model pretraining on huge text data sets with simple multitask and transfer learning techniques,” Facebook explains.

For the SuperGLUE benchmark, the consortium decided on tasks which meet four criteria:

1. Have varied formats.
2. Use more nuanced questions.
3. Are yet-to-be-solved using state-of-the-art methods.
4. Can be easily solved by people.

The new benchmark includes eight diverse and challenging tasks, including a Choice of Plausible Alternatives (COPA) causal reasoning task. The aforementioned task provides the system with the premise of a sentence and it must determine either the cause or effect of the premise from two possible choices. Humans have managed to achieve 100 percent accuracy on COPA while BERT achieves just 74 percent.

Across SuperGLUE’s tasks, RoBERTa is currently the leading NLP system and isn’t far behind the human baseline:

You can find a full breakdown of SuperGLUE and its various benchmarking tasks in a Facebook AI blog post here.

deepgeniusai.com/">AI & Big Data Expo events with upcoming shows in Silicon Valley, London, and Amsterdam to learn more. Co-located with the IoT Tech Expo, , & .

The post Esteemed consortium launch AI natural language processing benchmark appeared first on AI News.

Whereas mapping efforts from Google and Apple focus on navigating major cities, OpenStreetMap aims to ensure the smallest dirt paths are mapped.

Mapping the whole Earth is no small task: a sizeable local community helps, but it’s an impossible task manually both due to scale and often difficult to make out imagery.

In a Facebook blog post, mapping expert Dmitry Kuzhanov explained:

“Most modern algorithms, training sets, and techniques were invented to work for the areas with highly developed infrastructure.

In the developing world — for example, Africa, Southeast Asia, Latin America — where roads are not well-defined, maintained, or developed, even the best-trained human eye can struggle to identify and properly classify features.”

Over the past year and a half, Facebook mapped 300,000 miles of roads in Thailand for OpenStreetMap. These efforts resulted in the creation of a machine learning-based tool called RapiD which helps to lay computer-readable streets on top of satellite imagery.

Roads can be mapped even if they’re obscured by objects such as trees, while potential false positives like rivers are checked by volunteers. The model is based on a 34-layer deep neural network.

Martijn van Exel, a mapping leader in the OpenStreetMap community, said:

“The tool strikes a good balance between suggesting machine-generated features and manual mapping. It gives mappers the final say in what ends up in the map but helps just enough to both be useful and draw attention to undermapped places.

This is definitely going to be a key part of the future of OSM. We can never map the world, and keep it mapped, without assistance from machines. The trick is to find the sweet spot. OSM is a people project, and the map is a reflection of mappers’ interests, skills, biases, etc. That core tenet can never be lost, but it can and must travel along with new horizons in mapping.”

Facebook claims mapping the roads in Thailand by traditional means would have taken an additional three to five years.

RapiD is now available to help volunteers add and edit road data. The Map With AI service provides AI-generated maps in Afghanistan, Bangladesh, Indonesia, Mexico, Nigeria, Tanzania, and Uganda. Facebook says it will cover more countries in the future.

deepgeniusai.com/">AI & Big Data Expo events with upcoming shows in Silicon Valley, London, and Amsterdam to learn more. Co-located with the IoT Tech Expo, , & .

The post Facebook lends its AI powers to the OpenStreetMap community appeared first on AI News.

Facebook’s issue displaying images yesterday gave users an interesting look at how the social media giant’s AI analyses their uploads.

An outage yesterday meant Facebook users were unable to see uploaded images, providing a welcome respite from the usual mix of food and baby photos. In their place, however, was some interesting text.

Text in the placeholder where the image should have been displayed showed how Facebook’s AI automatically tagged the images.

Some of the aforementioned tags were understandable, like “one person, beard”. Other tags – such as a group of women standing together being tagged as “hoes” – were more questionable.

Facebook says it uses machine learning for tagging the images and reading the description to blind users. It’s unclear if there were “hoes” in terms of the farming tool in the image – but I don’t know how often even one is in a group photo, let alone plural.

Techies like many of our readers know Facebook is analysing photos to further understand each user, primarily for advertising purposes. Yesterday’s outage, however, will have opened more of the public’s eyes to how each of their uploads are being analysed and data extracted.

Shout-out to the person who uploaded a photo of their baby only for Facebook to categorise it as “Image may contain: dog”.

deepgeniusai.com/">AI & Big Data Expo events with upcoming shows in Silicon Valley, London, and Amsterdam to learn more. Co-located with the IoT Tech Expo, , & .

The post Facebook outage gave a glimpse at how its AI analyses images appeared first on AI News.

Last month, Facebook refused to remove a deepfake video of House Speaker Nancy Pelosi. Rather than making Pelosi appear to say things she never did, the video aimed to portray her as being intoxicated.

Deepfakes have the potential to spread misinformation and damage the reputation of individuals. Particularly in the world of politics and increasingly sophisticated state disinformation campaigns, it’s easy to imagine the danger they pose.

Pelosi later told California’s KQED: “I think they have proven — by not taking down something they know is false — that they were willing enablers of the Russian interference in our election.”

Governments and activists have called on social networks to assist in the detection and deletion of such videos. Now it seems that campaigners are targeting social network executives with deepfakes in a bid to make them take such videos more seriously.

In the deepfake of Zuckerberg, the Facebook CEO is portrayed to say: “Imagine this for a second: One man, with total control of billions of people’s stolen data, all their secrets, their lives, their futures.”

Fortunately for Zuckerberg, the deepfake is not malicious and most people will know it’s not real. The video both helps to make social networks assist more in tackling deepfakes while also raising public awareness.

Society is going through a period of serious change when it comes to knowing what’s real and fake. Many now question what they read because anyone could have written it and the issue of fake news is well-publicised. Most people, however, are still accustomed to believing that someone is saying what they are when they can see them doing it.

Zuckerberg’s deepfake was created Israeli startup Canny AI. The firm has also debuted fake videos with the likes of President Trump and Kim Kardashian as part of a commissioned art installation called Spectre that was on display at the Sheffield Doc/Fest in the UK.

deepgeniusai.com/">AI & Big Data Expo events with upcoming shows in Silicon Valley, London, and Amsterdam to learn more. Co-located with the IoT Tech Expo, , & .

The post Zuckerberg is deepfaked a month after Facebook refused to remove others appeared first on AI News.

The report highlights that – while gradually narrowing – the lack of diverse representation at major technology companies such as Microsoft, Google, and Facebook is causing AIs to cater more towards white males.

For example, at Facebook just 15 percent of the company’s AI staff are women. The problem is even more substantial at Google where just 10 percent are female.

Report authors Sarah Myers West, Meredith Whittaker and Kate Crawford wrote:

“To date, the diversity problems of the AI industry and the issues of bias in the systems it builds have tended to be considered separately.

We suggest that these are two versions of the same problem: issues of discrimination in the workforce and in system building are deeply intertwined.”

As artificial intelligence becomes used more across society, there’s a danger of some groups being left behind from its advantages while “reinforcing a narrow idea of the ‘normal’ person”.

The researchers highlight examples of where this is already happening:

• Amazon’s controversial Rekognition facial recognition AI struggled with dark-skin females in particular, although separate analysis has found other AIs also face such difficulties with non-white males.
• A résumé-scanning AI which relied on previous examples of successful applicants as a benchmark. The AI downgraded people who included “women’s” in their résumé or who attended women’s colleges.

AI is currently being deployed in few life-changing areas, but that’s rapidly changing. Law enforcement is already looking to use the technology for identifying criminals, even preemptively in some cases, and for making sentencing decisions – including whether someone should be granted bail.

“The use of AI systems for the classification, detection, and prediction of race and gender is in urgent need of re-evaluation,” the researchers noted. “The commercial deployment of these tools is cause for deep concern.”

deepgeniusai.com/">AI & Big Data Expo events with upcoming shows in Silicon Valley, London, and Amsterdam to learn more. Co-located with the IoT Tech Expo, , & .

The post Lack of STEM diversity is causing AI to have a ‘white male’ bias appeared first on AI News.