IBM's Deep Blue wasn't supposed to defeat Chess grandmaster Gary Kasparov when the two of them had their 1997 rematch. Computer experts of the time said machines would never beat us at strategy games because human ingenuity would always triumph over brute-force analysis. After Kasparov's loss, the experts didn't miss a beat. They said Chess was too easy and postulated that machines would never beat us at Go. Champion Lee Sedol's loss against DeepMind's AlphaGo proved them wrong there. Then the experts said AI would never beat us at games where strategy could be overcome by human creativity, such as poker.

Defenses against adversarial attacks, which in the context of AI refer to techniques that fool models through malicious input, are increasingly being broken by "defense-aware" attacks. In fact, most state-of-the-art methods claiming to detect adversarial attacks have been counteracted shortly after their publication. To break the cycle, researchers at the University of California, San Diego and Google Brain, including Turing Award winner Geoffrey Hinton, recently described in a preprint paper an approach that deflects attacks in the computer vision domain. Their framework either detects attacks accurately or, for undetected attacks, pressures the attackers to produce images that resemble the target class of images. The proposed architecture comprises (1) a network that classifies various input images from a data set and (2) a network that reconstructs the inputs conditioned on parameters of a predicted capsule.

A Bristol start-up building the next generation of microchips for artificial intelligence technology has raised $150m (£116m) from investors including Edinburgh's Baillie Gifford. Graphcore, founded in 2016 by Icera co-founder Nigel Toon, is now valued at almost $2bn having raised $450m from investors. Its new investors include Baillie Gifford, which has more than £200bn of assets under management, Mayfair Equity Partners and M&G Investments. Previous backers include Microsoft, BMW and Silicon Valley venture capital fund Sequoia. Graphcore has developed an intelligence processing unit, a microchip designed for artificial intelligence software.

Artificial intelligence is expected to increase economic output by $13 trillion in the coming decade. As per McKinsey's report, organisations that fully absorb this technology will double their cash flow in that time, while firms that don't, could see a 20% decline. The prepackaged solutions of the top cloud providers like Google Cloud help integrate AI into products. Cloud is now being utilised for navigating ships carrying cargo, thereby connecting online and offline retail markets. It is also assisting companies with ML services in fintech for fraud detection and many more.

Researchers in Japan have created a new kind of biodegradable, internet-connected piece of paper that's just one millimeter thick. Called nanopaper, the team believes it could change the way a wide variety of data is collected in nature. The project was developed at Osaka University's Institute of Scientific and Industrial Research (ISIR) by a team of scientists led by doctoral researcher Takaaki Kasuga. The nanopaper is just one-1,000th the thickness of pulp paper and it has a smooth surface that's similar to plastic in texture, according to a report in The Asahi Shimbun. In spite of its feel, the nanopaper is designed to be biodegradable and have a minimal impact on the environment and not require any expensive maintenance. In tests, the nanopaper had degraded by around 95 percent after 40 days in the ground, and its metal parts had been covered in non-toxic rust.

Any sufficiently advanced technology is indistinguishable from magic. In the world of artificial intelligence & machine learning (AI & ML), black- and white-box categorization of models and algorithms refers to their interpretability. That is, given a model trained to map data inputs to outputs (e.g. And just as the software testing dichotomy is high-level behavior vs low-level logic, only white-box AI methods can be readily interpreted to see the logic behind models' predictions. In recent years with machine learning taking over new industries and applications, where the number of users far outnumber experts that grok the models and algorithms, the conversation around interpretability has become an important one.

IBM has outlined principles to promote transparency -- and foster public trust -- in the way companies use artificial intelligence. The principles call on banks and other organizations to designate a lead AI official, own up to their use of the technology, explain it and test it for bias. Bankers say they're already on it. IBM unveiled the principles last month at Davos through its new IBM Policy Lab. The goal was to provide guidance for developing intelligent policy that will provide societal protections without stifling innovation.

NASA'S Planetary Defense Coordination Office uses the Catalina Sky Survey facility in Tucson,... [ ] Arizona, to catalog space objects Even in this age of high-speed data analysis, a keen human eye normally can't be beaten when poring over images of potential asteroidal impactors. But Artificial Intelligence (A.I.) could soon change all that. The El Segundo, Calif.-based Aerospace Corporation is now testing A.I. software designed to help astronomers speed up the process of identifying and tracking threatening Near-Earth Objects (NEOs). NASA's Planetary Defense Coordination Office already uses numerous telescopes to find and monitor NEOs that might have the potential to impact Earth. But the non-profit Aerospace Corporation's A.I. team is working with NASA on implementing software dubbed NEO AID (Near-Earth Object Artificial Intelligence Detection) to differentiate false positives from asteroids and comets that might be real threats.

And for in-house teams, labeling data can be the proverbial bottleneck, limiting a company's ability to quickly train and validate machine learning models. By its very definition, artificial intelligence refers to computer systems that can learn, reason, and act for themselves, but where does this intelligence come from? For decades, the collaborative intelligence of humans and machines has produced some of the world's leading technologies. And while there's nothing glamorous about the data being used to train today's AI applications, the role of data annotation in AI is nonetheless fascinating. Imagine reviewing hours of video footage – sorting through thousands of driving scenes, to label all of the vehicles that come into frame, and you've got data annotation.

A team of researchers at Leiden University in the Netherlands have developed a neural network called "Hazardous Object Identifier" that they say can predict if an asteroid is on a collision course with Earth. Their new AI singled out 11 asteroids that were not previously classified by NASA as hazardous, and which were larger than 100 meters in diameter -- big enough to explode with the force of hundreds of nuclear weapons if they impacted Earth, potentially leveling entire cities. They also focused on space rocks that could come within 4.7 million miles of Earth, as detailed in a paper published in the journal Astronomy & Astrophysics earlier this month. None are an imminent threat, however: not only are their chances of ever hitting Earth astronomically slim, but they are making their flyby between the years 2131 and 2923 -- hundreds of years from now. The team then reversed the simulation, simulating future Earth-impacting asteroids by flinging them away from Earth and tracking their exact locations and orbits.