Cloud Hosting

Geico, the nations second-biggest auto insurer, will try to speed up vehicle repairs for its policyholders by running photographs of damaged vehicles through artificial-intelligence software.

Berkshire Hathaway Inc. -owned Geico will offer the quick-estimate process in partnership with Tractable Ltd., said Alex Dalyac, chief executive and founder of the London-based technology firm. Tractable is among a number of specialists trying to help car insurers use artificial intelligence and other techniques to eliminate time-consuming hassles when customers file accident claims.

Financial terms of the partnership werent disclosed.

Todd Combs, Geicos chief executive, said in a written statement that Tractables technology is a way to obtain accurate estimates for policyholders and get drivers back on the road faster.

Geico is second in the private-passenger auto-insurance market, with a 13.6% share, according to the Insurance Information Institute. Geicos size as a part of Warren Buffetts Berkshire conglomerate means its moves are often followed by other car insurers, so its use of artificial intelligence in handling claims could become standard industry practice.

View original post here:
Geico to Use Artificial Intelligence to Speed Up Car Repairs - The Wall Street Journal

Read More..

An advanced artificial intelligence flight computer will help the UAEs lunar rover explore the surface of the Moon.

The navigation computer is being developed by Canadian space firm Mission Control Space Services.

It will recognise geological features as the Emirati rover, Rashid, drives around the unstable terrain of the lunar surface.

The computer will be installed on a Japanese lander that will take Rashid to the Moon next year, from where it will receive data from the rover. It will also send information back to Earth to be studied by scientists at the Mohammed bin Rashid Space Centre.

With the support of the Canadian Space Agency, Canadian scientists and engineers will be able to participate in near-term missions to the lunar surface, said Ewan Reid, president and chief executive of Mission Control.

Hind Bin Jarsh AlFalasi, 24, is the Emirati woman who has designed the logo for UAE's lunar mission. Reem Mohammed/The National

The Emirates Lunar Mission logo as revealed by Sheikh Hamdan bin Mohammed, Crown Prince of Dubai. It features the signature of Sheikh Rashid, the late ruler of Dubai. Courtesy: Sheikh Hamdan bin Mohammed Twitter

The logo will be featured on the Rashid Rover, which is slated for launch in 2022. Courtesy: MBRSC

Ms AlFalasi wanted to personalise the logo and show the mission's importance, so she added Sheikh Rashid's signature and inscribed it on top of the Moon. Reem Mohammed/The National

Only three countries have been able to land missions on the Moon so far. Courtesy: MBRSC

China is attempting another lunar mission. Its Chang'e 5 spacecraft has entered lunar orbit and aims to bring back rock and soil samples. UAE's lunar mission also aims to study lunar soil, as well as dust. Here, a Long March-5 rocket carrying the Chinese spacecraft lifts off on November 24. AP Photo/Mark Schiefelbein

Ms Al Falasi's career as a graphic designer has taken off. She will also be helping design the logo for the MBZ-Sat satellite, UAE's first fully in-house built spacecraft. Reem Mohammed/The National

The Mohammed bin Rashid Space Centre is carrying out the Emirates Lunar Mission. Reem Mohammed/The National

Rashid will explore the near side of the Moon, which offers a smoother surface with fewer craters, but the terrain is still unpredictable.

The four-wheeled rover can climb over an obstacle at a maximum height of 10 centimetres and descend a 20-degree slope.

But some basins on the near side of the moon are so steep that it would be impossible for the rover to climb out, were it to fall into one.

The team at the Mohammed bin Rashid Space Centre has shortlisted unexplored landing locations. The final decision will be based on an area that offers the most scientific value and security for the Arab worlds first lunar rover.

The navigation computer by Mission Control will include an AI application that will use deep-learning algorithms to recognise geological features in images captured by the rover.

This research will explore techniques for more advanced rover navigation, said Dr Melissa Battler, Mission Controls chief science officer.

By demonstrating this new technology on the moon, we will not only unlock potential autonomous decision-making capabilities for future rovers, but better support planetary-science missions going forward.

The company secured a $3.04 million fund from the Canadian Space Agencys Lunar Exploration Accelerator Programme, part of which will be used to develop the computer.

Japanese firm iSpace is building the Hakuto-Reboot lander that will deliver the rover to the Moon. Both will take off on board a Falcon 9 rocket from the Kennedy Space Centre in Florida late next year.

The Emirates Lunar Mission will also be provided with wired communication and power during the cruise phase and wireless communication on the lunar surface by iSpace.

Rashid will study the properties of lunar soil, the geology of the moon, dust movement and its photoelectron sheath for one lunar day about two weeks.

It will send back more than 1,000 images of the lunar surface.

Dr Hamad Al Marzooqi, project manager of the Emirates Lunar Mission, said it would be the first study of the photoelectron sheath.

It is a phenomenon that is created on the lunar surface due to the continuous bombardment of solar wind and cosmic rays," he said.

Sheikh Mohamed bin Zayed and Sheikh Mohammed bin Rashid personally thank staff from mission control in Dubai after Hope probe's successful orbit entry on February 9. The National

A man celebrates at an event at Burj Park in Dubai to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

People celebrate at an event at Burj Park to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

An event at Burj Park to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

People celebrate at an event at Burj Park to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

Guests arrive at the Burj Park event to mark the arrival of the Hope probe to Mars. Chris Whiteoak / The National

A guest attending the Burj Park event to mark the arrival of the Hope probe to Mars. Chris Whiteoak / The National

Burj Park was set up for people to watch the Hope probe attempt its Mars orbit insertion. Courtesy: UAE Government Twitter

UAE Mars Mission engineer, Hessa Al Matroushi, was interviewed at a Burj Park event to mark the arrival of the Hope probe to Mars. Chris Whiteoak / The National

Dr Thani Al Zeyoudi, Minister of State for Foreign Trade, attended the event at Burj Park to mark the arrival of the Hope probe to Mars. Chris Whiteoak / The National

Engineer Hessa Al Matroushi attended the event at Burj Park to mark the arrival of the Hope probe to Mars. Chris Whiteoak / The National

TV crews get ready at an event at Burj Park in Dubai to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

An event at Dubai's Burj Park to celebrate the Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

Guests arrive at an event at Burj Park to mark the Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

Guests and media arrive at an event at Burj Park to witness Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

Guests arrive at an event at Burj Park to mark the Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

Guests arrive at an event at Burj Park to mark the Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

Guests arrive at an event at Burj Park to mark the Hope probe's Mars orbit insertion attempt. Chris Whiteoak / The National

The Burj Khalifa lights up at an event at Burj Park in Dubai to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

The UAE Flag area on the Corniche in Abu Dhabi lights up in red to celebrate the success of the Hope probe going into orbit around Mars. Victor Besa / The National

The ADNOC Headquarters lights up in Abu Dhabi to celebrate the success of the Hope probe going into orbit around Mars. Victor Besa / The National

Sheikh Mohamed bin Zayed celebrates with Sheikh Mohammed bin Rashid at an event at Burj Park in Dubai to celebrate the Hope probe going into orbit around Mars. Chris Whiteoak / The National

People celebrate at an event at Burj Park in Dubai to mark the Hope probe going into orbit around Mars. Chris Whiteoak / The National

More:
UAEs lunar rover will use artificial intelligence to explore the Moon - The National

Read More..

Elevate your enterprise data technology and strategy at Transform 2021.

Adversarial machine learning, a technique that attempts to fool models with deceptive data, is a growing threat in the AI and machine learning research community. The most common reason is to cause a malfunction in a machine learning model. An adversarial attack might entail presenting a model with inaccurate or misrepresentative data as its training, or introducing maliciously designed data to deceive an already trained model.

As the U.S. National Security Commission on Artificial Intelligences 2019 interim report notes, a very small percentage of current AI research goes toward defending AI systems against adversarial efforts. Some systems already used in production could be vulnerable to attack. For example, by placing a few small stickers on the ground, researchers showed that they could cause a self-driving car to move into the opposite lane of traffic. Other studies have shown that making imperceptible changes to an image can trick a medical analysis system into classifying a benign mole as malignant, and that pieces of tape can deceive a computer vision system into wrongly classifying a stop signas a speed limit sign.

The increasing adoption of AI is likely to correlate with a rise in adversarial attacks. Its a never-ending arms race, but fortunately, effective approaches exist today to mitigate the worst of the attacks.

Attacks against AI models are often categorized along three primary axes influence on the classifier, the security violation, and their specificity and can be further subcategorized as white box or black box. In white box attacks, the attacker has access to the models parameters, while in black box attacks, the attacker has no access to these parameters.

An attack can influence the classifier i.e., the model by disrupting the model as it makes predictions, while a security violation involves supplying malicious data that gets classified as legitimate. A targeted attack attempts to allow a specific intrusion or disruption, or alternatively to create general mayhem.

Evasion attacks are the most prevalent type of attack, where data are modified to evade detection or to be classified as legitimate. Evasion doesnt involve influence over the data used to train a model, but it is comparable to the way spammers and hackers obfuscate the content of spam emails and malware. An example of evasion is image-based spam in which spam content is embedded within an attached image to evade analysis by anti-spam models. Another example is spoofing attacks against AI-powered biometric verification systems..

Poisoning, another attack type, is adversarial contamination of data. Machine learning systems are often retrained using data collected while theyre in operation, and an attacker can poison this data by injecting malicious samples that subsequently disrupt the retraining process. An adversary might input data during the training phase thats falsely labeled as harmless when its actually malicious. For example, large language models like OpenAIs GPT-3 can reveal sensitive, private information when fed certain words and phrases, research has shown.

Meanwhile, model stealing, also called model extraction, involves an adversary probing a black box machine learning system in order to either reconstruct the model or extract the data that it was trained on. This can cause issues when either the training data or the model itself is sensitive and confidential. For example, model stealing could be used to extract a proprietary stock-trading model, which the adversary could then use for their own financial gain.

Plenty of examples of adversarial attacks have been documented to date. One showed its possible to 3D-print a toy turtle with a texture that causes Googles object detection AI to classify it as a rifle, regardless of the angle from which the turtle is photographed. In another attack, a machine-tweaked image of a dog was shown to look like a cat to both computers and humans. So-called adversarial patterns on glasses or clothing have been designed to deceive facial recognition systems and license plate readers. And researchers have created adversarial audio inputs to disguise commands to intelligent assistants in benign-sounding audio.

In apaper published in April, researchers from Google and the University of California at Berkeley demonstrated that even the best forensic classifiers AI systems trained to distinguish between real and synthetic content are susceptible to adversarial attacks. Its a troubling, if not necessarily new, development for organizations attempting to productize fake media detectors, particularly considering the meteoric riseindeepfakecontent online.

One of the most infamous recent examples is Microsofts Tay, a Twitter chatbot programmed to learn to participate in conversation through interactions with other users. While Microsofts intention was that Tay would engage in casual and playful conversation, internet trolls noticed the system had insufficient filters and began feeding Tay profane and offensive tweets. The more these users engaged, the more offensive Tays tweets became, forcing Microsoft to shut the bot down just 16 hours after its launch.

As VentureBeat contributor Ben Dickson notes, recent years have seen a surge in the amount of research on adversarial attacks. In 2014, there were zero papers on adversarial machine learning submitted to the preprint server Arxiv.org, while in 2020, around 1,100 papers on adversarial examples and attacks were. Adversarial attacks and defense methods have also become a highlight of prominent conferences including NeurIPS, ICLR, DEF CON, Black Hat, and Usenix.

With the rise in interest in adversarial attacks and techniques to combat them, startups like Resistant AI are coming to the fore with products that ostensibly harden algorithms against adversaries. Beyond these new commercial solutions, emerging research holds promise for enterprises looking to invest in defenses against adversarial attacks.

One way to test machine learning models for robustness is with whats called a trojan attack, which involves modifying a model to respond to input triggers that cause it to infer an incorrect response. In an attempt to make these tests more repeatable and scalable, researchers at Johns Hopkins University developed a framework dubbed TrojAI, a set of tools that generate triggered data sets and associated models with trojans. They say that itll enable researchers to understand the effects of various data set configurations on the generated trojaned models and help to comprehensively test new trojan detection methods to harden models.

The Johns Hopkins team is far from the only one tackling the challenge of adversarial attacks in machine learning. In February, Google researchers released apaper describing a framework that either detects attacks or pressures the attackers to produce images that resemble the target class of images. Baidu, Microsoft, IBM, and Salesforce offer toolboxes Advbox, Counterfit, Adversarial Robustness Toolbox, and Robustness Gym for generating adversarial examples that can fool models in frameworks like MxNet, Keras, Facebooks PyTorch and Caffe2, Googles TensorFlow, and Baidus PaddlePaddle. And MITs Computer Science and Artificial Intelligence Laboratory recently released a tool called TextFoolerthat generates adversarial text to strengthen natural language models.

More recently, Microsoft, the nonprofit Mitre Corporation, and 11 organizations including IBM, Nvidia, Airbus, and Bosch releasedtheAdversarial ML Threat Matrix, an industry-focused open framework designed to help security analysts to detect, respond to, and remediate threats against machine learning systems. Microsoft says it worked with Mitre to build a schema that organizes the approaches malicious actors employ in subverting machine learning models, bolstering monitoring strategies around organizations mission-critical systems.

The future might bring outside-the-box approaches, including several inspired by neuroscience. For example, researchers at MIT and MIT-IBM Watson AI Lab have found that directly mapping the features of the mammalian visual cortex onto deep neural networks creates AI systems that are more robust to adversarial attacks. While adversarial AI is likely to become a never-ending arms race, these sorts of solutions instill hope that attackers wont always have the upper hand and that biological intelligence still has a lot of untapped potential.

See the rest here:
Adversarial attacks in machine learning: What they are and how to stop them - VentureBeat

Read More..

Job loses caused by automation may grab the bulk of the headlines, but more of us may be affected by changes to recruitment and worker surveillance, writes Colin Gavaghan, director of the Centre for Law and Policy in Emerging Technologies

Until recently, a question such as that in the headline has led immediately to discussions about how many jobs will be lost to the technological revolution of artificial intelligence. Over the past few years, though, more of us have started looking at some other aspects of this question. Such as: for those of us still in work, how will things change? What will it be like to work alongside, or under, AI and robots? Or to have decisions about whether were hired, fired or promoted made by algorithms?

Those are some of the questions our multi-disciplinary team at Otago University, funded by the New Zealand Law Foundation, have been trying to answer. Last week, we set out our findings in a new report.

Theres a danger of getting a bit too Black Mirror about these sorts of things, of always seeing the most dystopian possibilities from any new technology. Thats a trap weve tried hard to avoid, because there really are potential benefits in the use of this sort of technology. For one thing, its possible that AI and robots could make some workplaces safer. ACC recently invested in Kiwi robotics company Robotics Plus, for example, whose products are intended to reduce the risk of accidents at ports, forestry sites and sawmills.

Of course, workplace automation can also increase danger. Weve already seen examples of workplace robots causing fatalities. One of our suggestions is that New Zealands work safety rules need to catch up with the sort of robots were likely to be working alongside in the future fencing them off from human workers and providing an emergency off-switch isnt going to be the answer for cobots that are designed to work with and around us.

Physical injuries from robots may present the most visceral image of the risks of workplace automation. Luckily, theyre probably likely to be fairly rare. Far more people, we think, will be affected by algorithmic management the growing range of techniques used to allocate shifts, direct workers and monitor performance.

As with workplace robots, theres potential here for the technology to improve things for workers. One report talked about how it could benefit workers by giving clearer advance notice of when shifts will be and making it easier to swap and change them. Theres no guarantee, though, that algorithmic management tools will be used to benefit workers. Our earlier warning aside, its hard not to feel just a bit Black Mirror when seeing images of Amazon warehouses where workers are micro-managed to an extent beyond the wildest dreams of Ford or Taylor.

An Amazon fulfillment centre in Illinois, USA (Photo: Scott Olson)

A particular concern thats grown during the Covid crisis is the apparently increasing prevalence of workplace surveillance. While by no means a new phenomenon, AI technologies could offer employers the opportunity to monitor their workers more closely and ubiquitously than ever before.

Of course, not all employers will treat their workers like drones. But workplace protection rules dont exist for the good employers. If we want to avoid the soul-crushing erosion of privacy, autonomy and dignity that could accompany the worst abuses of this technology, we think those rules will need to be tightened in various ways.

Concerns about AI in the workplace dont start with algorithmic management, though. A lot of them start before the employment relationship even begins. Increasingly, AI technology is being used in recruitment: from targeted job adverts, to shortlisting of applicants, even to the interview stage, where companies like Hirevue provide algorithms to analyse recorded interviews with candidates.

The use of algorithms in hiring poses a serious risk of reinforcing bias, or of rendering existing bias less visible. Most obviously, theres a risk that algorithms will base their profiles of a good fit for a particular role on the history of people whove occupied that role before. If those people happen to have been overwhelming white, male and middle class well, its not hard to guess how that will probably go. Also, affective recognition software thats been trained on mostly white, neurotypical people could make unfair adverse judgments about people who dont fit into those categories, even if they score highly in the sorts of attributes that really matter. (Hirevue recently stopped using visual analysis for their assessment models, but since these sorts of platforms will obviously have to rely on inferences from something maybe voice inflection or word choices questions about cultural, class or neurodiversity awareness remain.)

But doesnt New Zealand already have laws protecting us against workplace hazards, privacy violations and discrimination? It does indeed. Like almost every other new technology, workplace AI isnt emerging into a legal vacuum. Unfortunately, some of those laws were designed for a different time, which can lead to what tech lawyers call regulatory disconnection when theres a major change to the technologys form or use. For instance, the current rules around workplace robots seem to assume that they can be fenced off from human workers, whereas the cobots that are now coming into use will be working in close proximity to humans.

In other cases, the law seems fine, but the problem is spotting when the technology violates it. Our Human Rights Act prevents discrimination on a whole bunch of grounds, including sex, race and disability, but that wont be much help to someone who has no way of knowing why the algorithm has declined them. It may even be that employers themselves wont know who has been screened out at an early stage, or on what grounds.

As we argue, though, it doesnt have to be that way. Just as workplace robots could reduce injuries and fatalities, so could algorithmic auditing software help to detect and reduce bias in recruitment, promotion, etc. Its not as though humans are perfect at this! Maybe AI could make things better. What we cant do, though, is complacently assume that it will do so.

In April, the EU Commission published a draft law for Europe which would require scrutiny and transparency for certain uses of AI technology. That would include a range of functions related to employment, such as recruitment, promotion and termination; task allocation; and monitoring and evaluating performance. Last year, New York City Council introduced a bill that would provide for algorithmic hiring tools be audited for bias, and their use disclosed to candidates.

Our report calls for New Zealand to take the same kinds of steps. For instance, we propose that consideration should be given to following New Yorks example, and requiring manufacturers of hiring tools to ensure those tools include functionality for bias auditing, so that client companies can readily perform the relevant audits.

Algorithmic impact assessments looking at matters like privacy and worker safety and wellbeing should be conducted before any algorithm is used in high stakes contexts. And weve suggested that there should be important roles for the Office of the Privacy Commissioner and Worksafe NZ in overseeing surveillance technologies.

We think these steps would go some way to ensuring that New Zealand businesses and workers (actual and prospective) could enjoy the benefits of these technologies, while being protected from the worst of the risks.

Our report isnt a prediction about what the future workplace will look like when AI and robots are a regular part of it. How things turn out depends substantially on the sorts of choices we make about how to use these technologies. And were not proposing that we need fear the future or rage against the machines. But we do think we should be keeping a close watchful eye on them. Because you can bet theyll be keeping an eye on us.

Subscribe to Rec Room a weekly newsletter delivering The Spinoffs latest videos, podcasts and other recommendations straight to your inbox.

Link:
How will artificial intelligence change the way we work? Theres good and bad news - The Spinoff

Read More..