US company Lyft has filed plans with the regulator in Washington for a flotation as it races for funding against Uber, the rival ride-hailing company. In a statement on Thursday, Lyft announced it had submitted a draft registration statement with the US Securities and Exchange Commission (SEC), in a move which sets it up to be one of the first large tech flotations of 2019. The number of shares on offer and the price range for the proposed offering have not yet been determined, Lyft said. The company was valued at $15.1bn (£11.8bn) at its latest funding round in June, when investors led by Fidelity Investments poured $600m into the firm. Uber, by contrast, has been valued at $120bn in recent reports.

Artificial intelligence has made major strides in the past few years, but those rapid advances are now raising some big ethical conundrums. Chief among them is the way machine learning can identify people's faces in photos and video footage with great accuracy. This might let you unlock your phone with a smile, but it also means that governments and big corporations have been given a powerful new surveillance tool. A new report from the AI Now Institute (large PDF), an influential research institute based in New York, has just identified facial recognition as a key challenge for society and policymakers. The speed at which facial recognition has grown comes down to the rapid development of a type of machine learning known as deep learning.

Donald Trump's commitment to "remain[ing] a steadfast partner of Saudi Arabia," despite the regime's gruesome torture and murder of journalist Jamal Khashoggi in Turkey, is clearly symptomatic of the malignantly self-serving nature of US foreign policy, which has long propped up dictatorships and enabled atrocities around the world for the sake of profit and power. However, many of Trump's most vocal critics on the Saudi file show signs of an equally dangerous pathological condition: a profound historical amnesia that permits some of the most prominent proponents of the US' own torturous and murderous policies to now parade as champions of human rights, without any apparent sense of irony. Obama-era CIA Director John Brennan, for instance, has insisted that "the US should never turn a blind eye to this sort of inhumanity [referring to the murder of Khashoggi] … because this is a nation that remains faithful to its values" - a curiously self-righteous stance for a man who not only repeatedly turned a blind eye to the inhumanity of past and present CIA practices such as extraordinary rendition, torture, and drone assassination, but actively defended and (in the case of drone use) expanded them. Senate Majority Leader Mitch McConnell decried the brutal murder of Khashoggi as "completely abhorrent to everything the United States holds dear and stands for in the world". Yet he praised another perpetrator of abhorrent deeds, CIA "black site" torture prison manager Gina Haspel, as an "excellent choice" for Director of the CIA.

After a hellish year of tech scandals, even government-averse executives have started professing their openness to legislation. But Microsoft president Brad Smith took it one step further on Thursday, asking governments to regulate the use of facial-recognition technology to ensure it does not invade personal privacy or become a tool for discrimination or surveillance. Tech companies are often forced to choose between social responsibility and profits, but the consequences of facial recognition are too dire for business as usual, Smith said. "We believe that the only way to protect against this race to the bottom is to build a floor of responsibility that supports healthy market competition," he said in a speech at the Brookings Institution. "We must ensure that the year 2024 doesn't look like a page from the novel 1984."

One of the main drawbacks of deep neural networks, like many other classifiers, is their vulnerability to adversarial attacks. An important reason for their vulnerability is assigning high confidence to regions with few or even no feature points. By feature points, we mean a nonlinear transformation of the input space extracting a meaningful representation of the input data. On the other hand, deep-RBF networks assign high confidence only to the regions containing enough feature points, but they have been discounted due to the widely-held belief that they have the vanishing gradient problem. In this paper, we revisit the deep-RBF networks by first giving a general formulation for them, and then proposing a family of cost functions thereof inspired by metric learning. In the proposed deep-RBF learning algorithm, the vanishing gradient problem does not occur. We make these networks robust to adversarial attack by adding the reject option to their output layer. Through several experiments on the MNIST dataset, we demonstrate that our proposed method not only achieves significant classification accuracy but is also very resistant to various adversarial attacks.

The study of fairness in intelligent decision systems has mostly ignored long-term influence on the underlying population. Yet fairness considerations (e.g. affirmative action) have often the implicit goal of achieving balance among groups within the population. The most basic notion of balance is eventual equality between the qualifications of the groups. How can we incorporate influence dynamics in decision making? How well do dynamics-oblivious fairness policies fare in terms of reaching equality? In this paper, we propose a simple yet revealing model that encompasses (1) a selection process where an institution chooses from multiple groups according to their qualifications so as to maximize an institutional utility and (2) dynamics that govern the evolution of the groups' qualifications according to the imposed policies. We focus on demographic parity as the formalism of affirmative action. We then give conditions under which an unconstrained policy reaches equality on its own. In this case, surprisingly, imposing demographic parity may break equality. When it doesn't, one would expect the additional constraint to reduce utility, however, we show that utility may in fact increase. In more realistic scenarios, unconstrained policies do not lead to equality. In such cases, we show that although imposing demographic parity may remedy it, there is a danger that groups settle at a worse set of qualifications. As a silver lining, we also identify when the constraint not only leads to equality, but also improves all groups. This gives quantifiable insight into both sides of the mismatch hypothesis. These cases and trade-offs are instrumental in determining when and how imposing demographic parity can be beneficial in selection processes, both for the institution and for society on the long run.

As machine learning models continue to increase in complexity, collecting large hand-labeled training sets has become one of the biggest roadblocks in practice. Instead, weaker forms of supervision that provide noisier but cheaper labels are often used. However, these weak supervision sources have diverse and unknown accuracies, may output correlated labels, and may label different tasks or apply at different levels of granularity. We propose a framework for integrating and modeling such weak supervision sources by viewing them as labeling different related sub-tasks of a problem, which we refer to as the multi-task weak supervision setting. We show that by solving a matrix completion-style problem, we can recover the accuracies of these multi-task sources given their dependency structure, but without any labeled data, leading to higher-quality supervision for training an end model. Theoretically, we show that the generalization error of models trained with this approach improves with the number of unlabeled data points, and characterize the scaling with respect to the task and dependency structures. On three fine-grained classification problems, we show that our approach leads to average gains of 20.2 points in accuracy over a traditional supervised approach, 6.8 points over a majority vote baseline, and 4.1 points over a previously proposed weak supervision method that models tasks separately.

As artificial intelligence (AI) systems become increasingly ubiquitous, the topic of AI governance for ethical decision-making by AI has captured public imagination. Within the AI research community, this topic remains less familiar to many researchers. In this paper, we complement existing surveys, which largely focused on the psychological, social and legal discussions of the topic, with an analysis of recent advances in technical solutions for AI governance. By reviewing publications in leading AI conferences including AAAI, AAMAS, ECAI and IJCAI, we propose a taxonomy which divides the field into four areas: 1) exploring ethical dilemmas; 2) individual ethical decision frameworks; 3) collective ethical decision frameworks; and 4) ethics in human-AI interactions. We highlight the intuitions and key techniques used in each approach, and discuss promising future research directions towards successful integration of ethical AI systems into human societies.

Uber Technologies Inc's UBER.UL is plotting the return of its self-driving cars, but the company known for its hard-charging style is taking a much more conservative approach as it tries to recover from a fatal accident that upended its autonomous vehicle program. After it receives approval from the state of Pennsylvania, Uber plans to begin driving'a handful' of cars on a mile loop between two company offices in Pittsburgh, where Uber first debuted its autonomous vehicles in 2016, company spokeswoman Sarah Abboud said. The new diminutive launch is a dramatic downsize from Uber's previous operation. The company in the past unleashed its fleet in autonomous mode on public roads at high speeds, after dark, in areas crowded with pedestrians and with a single backup driver in the front seat. This time, the cars will not operate at night or in wet weather, and will not exceed speeds of 25 miles per hour, Abboud said.

A new weapon in the fight to protect species from extinction has been launched, in the form of a a NASA space laser satellite that will monitor Earth's forests. Gedi - pronounced Jedi, a reference to Star Wars - will use lidar technology as it soars over the dark and light side of the planet. The device, which is the size of a fridge, will help scientists understand why forests are declining and experts hope this will help them reverse the trend. By working out the spread and density of tropical and temperate forests, scientists can work out which habitats and therefore species are most at risk. A NASA space laser satellite that is the size of a fridge will map and monitor Earth's forests to protect at risk species from extinction.