Businesses are crawling social media, email and internal instant messaging services for employees making sexist or bullying comments in an attempt to root out troublesome behaviour and avoid lawsuits. Fama, a California start-up which claims to have 120 clients including Fortune 500 companies, said it is helping businesses weed out individuals likely to cause a rift among workers and expose the business to costly lawsuits. Its artificial intelligence-powered snooping software identified 82,900 instances of misogyny, 40,200 instances of bigotry, 677 insinuations of violence and 589 instances of criminal behaviour in 2018. Fama claims to scan 15,000 workers per month, including a number in the...

The increasing adoption of machine learning tools has led to calls for accountability via model interpretability. But what does it mean for a machine learning model to be interpretable by humans, and how can this be assessed? We focus on two definitions of interpretability that have been introduced in the machine learning literature: simulatability (a user's ability to run a model on a given input) and "what if" local explainability (a user's ability to correctly indicate the outcome to a model under local changes to the input). Through a user study with 1000 participants, we test whether humans perform well on tasks that mimic the definitions of simulatability and "what if" local explainability on models that are typically considered locally interpretable. We find evidence consistent with the common intuition that decision trees and logistic regression models are interpretable and are more interpretable than neural networks. We propose a metric - the runtime operation count on the simulatability task - to indicate the relative interpretability of models and show that as the number of operations increases the users' accuracy on the local interpretability tasks decreases.

The report also describes a more effective form of "digital authoritarianism" that is leading the assault on freedom of speech. China, in particular, is actively exporting its approach to internet censorship and surveillance around the world. In an earlier report, "Freedom on the Net," Freedom House outlined how China is offering training sessions and study trips, as well as advanced equipment that takes advantage of artificial intelligence and facial recognition technologies, to monitor internet activity.

Apple could be planning to introduce an emojii version of its Siri virtual assistant, according to a new patent application from the tech giant. The patent request, filed with the US Patent and Trademark Office, describes an emoji-based avatar for a smart home speaker that can adapt to a user's mood. Though not mentioned by name in the patent, the description of the smart speaker accurately resembles that of the Apple HomePod. Apple's patent application describes a "humanistic avatar, a simplified graphical representation of a digital assistant such as an emoji-based avatar" – essentially a cartoon version of Siri. Depending on what request is made through the smart speaker, the emoji assistant would be able to react appropriately.

Faced with mounting criticism of its "Rekognition" system, Amazon has come out in favor of legislating facial recognition technology. In a blog post, the company has revealed its "proposed guidelines" for the responsible use of the tech that it hopes policymakers in the US and worldwide will consider when drafting new laws. Amazon's five-step rulebook essentially calls for use of the tech to be governed by current laws, including those that protect civil rights. It also urges human oversight when facial recognition is used by law enforcement and recommends a 99 percent confidence score threshold for identification, adding that the tech should not be the "sole determinant" in an investigation. It calls for law enforcement to release regular transparency reports on their use of the systems.

I interviewed Chris Priebe, CEO of Two Hat, which recently released an artificial intelligence model that moderates user-generated reports in real time on social media to prevent abuse, hate speech, and other NSFW content. Chris Priebe has over 20 years' experience with fostering healthy online communities and is deeply passionate about making the internet a safer place. Chris was the lead developer on the safety and security elements for Club Penguin, which was acquired by Disney and grew to over 300 million users. Chris founded Two Hat in 2012 and began coding Community Sift, a content moderation solution for social platforms that detects and filters high-risk content like bullying, hate speech, and grooming. Today, some of the biggest social platforms in the world use Community Sift to protect users from abusive and unwanted chat, images, and videos.

In Washington County, Oregon, sheriff's deputies use a mobile app to send photos of suspects to Amazon's cloud computing service. The e-commerce giant's algorithms check those faces against a database of tens of thousands of mugshots, using Amazon's Rekognition image analysis service. Such use of facial recognition by law enforcement is essentially unregulated. But some developers of the technology want to change that. In a blog post Thursday, Amazon asked Congress to put some rules around the use of the technology, echoing a call by Microsoft in December.

In many applications, we want to predict and estimate the effect of a covariate on survival timeof interests. Examples include treatment, surgical procedure, or immunization on survival time of patients, who for example, could be individuals who have metastatic breast cancer, military casualties suffering from various injuries, or survival time of infectious diseases.Classically, most datasets have been too small to meaningfully examine the heterogeneity of the data beyond dividing them into a few subpopulations. In the past few years, however, there has been an explosion of experimental settings where it is potentially feasible to explore heterogeneity to its full extent. An impediment to exploring heterogeneous effects is the fear that scientists with two opposite agendas could hypothetically string together two opposite but coherent results by searching through many different possible models and then reporting only the very extreme ones - highlighting solely spurious results (Olken, 2015). Thus, protocols for clinical trials must specify in advance the pre-analysis plans and then learn from the data.

Consequential decisions are increasingly informed by sophisticated data-driven predictive models. For accurate predictive models, deterministic threshold rules have been shown to be optimal in terms of utility, even under a variety of fairness constraints. However, consistently learning accurate models requires access to ground truth data. Unfortunately, in practice, some data can only be observed if a certain decision was taken. Thus, collected data always depends on potentially imperfect historical decision policies. As a result, learned deterministic threshold rules are often suboptimal. We address the above question from the perspective of sequential policy learning. We first show that, if decisions are taken by a faulty deterministic policy, the observed outcomes under this policy are insufficient to improve it. We then describe how this undesirable behavior can be avoided using stochastic policies. Finally, we introduce a practical gradient-based algorithm to learn stochastic policies that effectively leverage the outcomes of decisions to improve over time. Experiments on both synthetic and real-world data illustrate our theoretical results and show the efficacy of our proposed algorithm.

I experienced this firsthand, when I was a graduate student at MIT in 2015 and discovered that some facial analysis software couldn't detect my dark-skinned face until I put on a white mask. These systems are often trained on images of predominantly light-skinned men. And so, I decided to share my experience of the coded gaze, the bias in artificial intelligence that can lead to discriminatory or exclusionary practices. Altering myself to fit the norm--in this case better represented by a white mask than my actual face--led me to realize the impact of the exclusion overhead, a term I coined to describe the cost of systems that don't take into account the diversity of humanity. How much does a person have to change themselves to function with technological systems that increasingly govern our lives? We often assume machines are neutral, but they aren't.