Eat too much and there won't be grass for anyone. In an essay written in 1833, the British economist William Forster Lloyd made a profound observation using the example of cattle grazing. Lloyd described a hypothetical scenario involving herders who share a pasture, and individually decide how many of their animals would graze there. If few herders exercised restraint, overgrazing would occur, reducing the pasture's future usefulness and eventually hurting everybody. The sinister beauty of this example is that the rational course of action is to behave selfishly.

This'quadrupedal' vehicle may look like a smart shopping trolley ready for a supermarket dash in some distant interstellar community. But, in fact, it's a full-size, robotic walking car, which Hyundai believes may be helpful in rescue zones when normal vehicles, even the most robust 4x4s, just can't hack it. It's called the'Elevate' and by blending technology found in modern electric cars with advanced robotics, it can climb up 5ft walls, straddle a 5ft hole and step across piles of debris, thanks to the addition of four fully articulated robotic legs – and all the while keeping its passengers completely level. The idea is that the Elevate could be driven by first responders to a disaster location, just like a traditional electric car, but then when the terrain became impassable it could use its highly dexterous legs to move in any direction. It can walk at 3mph and the legs are powered by the same battery that drives the car's motor.

A growing number of lenders are using artificial intelligence to digest growing volumes of data and find relationships between variables to determine creditworthiness. Last year, subprime auto lender Prestige Financial Services started working with artificial intelligence (AI) software developer ZestFinance to analyze about 2,700 borrower characteristics, instead of the several dozen the lender had on its risk-assessment scorecard. Draper, UT-based Prestige is among a growing number of lenders that view AI as a tool that can digest increasing volumes of data and find relationships between variables to determine creditworthiness. Prestige and ZestFinance developed a machine learning system that allowed Prestige to consider factors such as when a bankruptcy happened, previous car-payment records, and time spent living at a current residence. Said ZestFinance's Douglas Merrill, "If you're building an AI model, you can have hundreds or thousands" of such indicators, including whether people have defaulted on rent payments or cellphone bills.

Algorithms are increasingly being used to make ethical decisions. Perhaps the best example of this is a high-tech take on the ethical dilemma known as the trolley problem: if a self-driving car cannot stop itself from killing one of two pedestrians, how should the car's control software choose who live and who dies? In reality, this conundrum isn't a very realistic depiction of how self-driving cars behave. But many other systems that are already here or not far off will have to make all sorts of real ethical trade-offs. Assessment tools currently used in the criminal justice system must consider risks to society against harms to individual defendants; autonomous weapons will need to weigh the lives of soldiers against those of civilians.

Over the past week, the #2009vs2019 meme challenge, alternately known as the #10yearchallenge and #HowHardDidAgeHitYou, has become the latest social media trend ripe for think piece fodder. While the challenge inspired a host of discussions about social media narcissism and gendered norms, author and consultant Kate O'Neill put her own spin on the meme in a tweet raising the privacy implications of posting age-separated photos of oneself on Facebook. The post generated enough buzz and discussion on Twitter that O'Neill expanded it into an article in Wired, in which she argued that Facebook or another data-hungry entity could exploit the meme to train facial recognition algorithms to better handle age-related characteristics and age progression predictions. She noted that the clear labeling of the year in which the pictures were taken, along with the volume of pictures explicitly age-separated by a set amount of time, could be quite valuable to a company like Facebook. "In other words, thanks to this meme, there's now a very large data set of carefully curated photos of people from roughly 10 years ago and now," O'Neill wrote.

It gets knocked down, but it gets up again. This dog-like robot learns to explore all the ways to stand up after falling over – or being shoved, as they often are during testing – one of the toughest tests for four-legged robots to pass. ANYmal is about the size of a large dog, standing 70 centimetres high and weighing 35 kilograms. Its has 12 moving parts that must be coordinated to walk, run or right the robot after it falls over. Modelling all those points and the positions they could potentially take in various landscapes and at different speeds would take weeks for a human to input, says Jemin Hwangbo at ETH Zurich in Switzerland, who led the study.

In 2017, a poker bot called Libratus made headlines when it roundly defeated four top human players at no-limit Texas Hold'Em. Now, Libratus's technology is being adapted to take on opponents of a different kind--in service of the US military. Libratus--Latin for balanced--was created by researchers from Carnegie Mellon University to test ideas for automated decision-making based on game theory. Early last year, the professor who led the project, Tuomas Sandholm, founded a startup called Strategy Robot to adapt his lab's game-playing technology for government use, such as in wargames and simulations used to explore military strategy and planning. Late in August, public records show, the company received a two-year contract of up to $10 million with the US Army.

Much of recent machine learning has focused on deep learning, in which neural network weights are trained through variants of stochastic gradient descent. An alternative approach comes from the field of neuroevolution, which harnesses evolutionary algorithms to optimize neural networks, inspired by the fact that natural brains themselves are the products of an evolutionary process. Neuroevolution enables important capabilities that are typically unavailable to gradient-based approaches, including learning neural network building blocks (for example activation functions), hyperparameters, architectures and even the algorithms for learning themselves. Neuroevolution also differs from deep learning (and deep reinforcement learning) by maintaining a population of solutions during search, enabling extreme exploration and massive parallelization. Finally, because neuroevolution research has (until recently) developed largely in isolation from gradient-based neural network research, it has developed many unique and effective techniques that should be effective in other machine learning areas too.

Federal regulators have announced plans to allow drone operators to fly their unmanned aerial vehicles over populated areas and at night. A Wing Hummingbird drone from Project Wing arrives and sets down its package at a delivery location in Blacksburg, Va., last year. Federal regulators have announced plans to allow drone operators to fly their unmanned aerial vehicles over populated areas and at night. A Wing Hummingbird drone from Project Wing arrives and sets down its package at a delivery location in Blacksburg, Va., last year. Package delivery by drone is one small step closer to reality today.