In the United States and Europe, the debate in the artificial intelligence community has focused on the unconscious biases of those designing the technology. Recent tests showed facial recognition systems made by companies like I.B.M. and Amazon were less accurate at identifying the features of darker-skinned people. China's efforts raise starker issues. While facial recognition technology uses aspects like skin tone and face shapes to sort images in photos or videos, it must be told by humans to categorize people based on social definitions of race or ethnicity. Chinese police, with the help of the start-ups, have done that.

Ask a layman about artificial intelligence and they might point to sci-fi villains such as HAL from 2001: A Space Odyssey or the Terminator. But the co-founders of the AI Now Institute, Meredith Whittaker and Kate Crawford, want to change the conversation. Instead of talking about far-flung super-intelligent AI, they argued on the latest episode of Recode Decode, we should be talking about the ways AI is affecting people right now, in everything from education to policing to hiring. Rather than killer robots, you should be concerned about what happens to your résumé when it hits a program like the one Amazon tried to build. "They took two years to design, essentially, an AI automatic résumé scanner," Crawford said. "And they found that it was so biased against any female applicant that if you even had the word'woman' on your résumé that it went to the bottom of the pile." That's a classic example of what Crawford calls "dirty data." Even though people think of algorithms as being ...

Eric Horvitz is a technical fellow and director at Microsoft Research Labs. A recipient of the Feigenbaum and the Allen Newell Prizes for contributions to artificial intelligence (AI), he is also on the US President's Council of Advisors on Science and Technology, Defense Advanced Research Projects Agency, and the Allen Institute for Artificial Intelligence. He is also part of the standing committee of Stanford University's One Hundred Year Study on Artificial Intelligence. Horvitz, who comes at least once a year to the country to interact with the India labs team, spoke about his work at Microsoft Research. He also shared his thoughts on the benefits and fear of AI, and attempts to address the bias in algorithms.

The Chinese government is using facial-recognition software to "track and control" a predominantly Muslim minority group, according to a disturbing new report from The New York Times. The Chinese government has reportedly integrated artificial intelligence into its security cameras to identify the Uighurs and appears to be using the information to monitor the persecuted group. The report, based on the accounts of whistleblowers familiar with the systems and a review of databases used by the government and law enforcement, suggests the authoritarian country has opened up a new frontier in the use of A.I. for racist social control--and raises the discomfiting possibility that other governments could adopt similar practices. Two people familiar with the matter told the Times that police in the Chinese city of Sanmenxia screened whether residents were Uighurs 500,000 times in a single month. Documents provided to the paper reportedly show demand for the technology is ballooning: more than 20 departments in 16 provinces sought access to the camera system, in one case writing that it "should support facial recognition to identify Uighur/non-Uighur attributes." This, experts say, is more than enough to raise red flags.

The operator of the tsunami-wrecked Fukushima nuclear plant began removing fuel Monday from a cooling pool at one of three reactors that melted down in the 2011 disaster, a milestone in what will be a decades-long process to decommission the facility. Tokyo Electric Power Co. said workers started removing the first of 566 used and unused fuel units stored in the pool at Unit 3. The fuel units in the pool located high up in reactor buildings are intact despite the disaster, but the pools are not enclosed, so removing the units to safer ground is crucial to avoid disaster in case of another major earthquake similar to the one that caused the 2011 tsunami. TEPCO says the removal at Unit 3 will take two years, followed by the two other reactors, where about 1,000 fuel units remain in the storage pools. Removing fuel units from the cooling pools comes ahead of the real challenge of removing melted fuel from inside the reactors, but details of how that might be done are still largely unknown. Removing the fuel in the cooling pools was delayed more than four years by mishaps, high radiation and radioactive debris from an explosion that occurred at the time of the reactor meltdowns, underscoring the difficulties that remain.

The operator of the tsunami-wrecked Fukushima nuclear plant has begun removing fuel from a cooling pool at one of three reactors that melted down in the 2011 disaster, a milestone in the decades-long process to decommission the plant. Tokyo Electric Power Co (Tepco) said on Monday that workers started removing the first of 566 used and unused fuel units stored in the pool at Unit 3. The fuel units in the pool located high up in reactor buildings are intact despite the disaster, but the pools are not enclosed so removing the units to safer ground is crucial to avoid disaster in case of another major quake. Tepco said the removal at Unit 3 would take two years, followed by the two other reactors. The step comes ahead of the real challenge of removing melted fuel from inside the reactors, but details of how that might be done are still largely unknown. Removing the fuel in the cooling pools was delayed five years by mishaps, high radiation and radioactive debris from an explosion that occurred at the time of the reactor meltdown, underscoring the difficulties that remain.

Vikram Mahidhar reminds us all that AI is only as good as the humans supervising it and programming it. The biases and artefacts that come out of the processing are reflective of the biases programmed in at the beginning. A program trained to recognise totalled car bodies for insurance purposes, for example, will need close supervision of its decision-making outputs, for regulatory and consumer confidence and acceptance of the decision. There is a call and a growth in a new class of AI--one that is explainable, and that builds trust by providing evidence. Vikram also reminds us that a governance strategy is key to engendering trust in our organisation, processes and people.

In predictive process analytics, current and historical process data in event logs are used to predict future. E.g., to predict the next activity or how long a process will still require to complete. Recurrent neural networks (RNN) and its subclasses have been demonstrated to be well suited for creating prediction models. Thus far, event attributes have not been fully utilized in these models. The biggest challenge in exploiting them in prediction models is the potentially large amount of event attributes and attribute values. We present a novel clustering technique which allows for trade-offs between prediction accuracy and the time needed for model training and prediction. As an additional finding, we also found that this clustering method combined with having raw event attribute values provides even better prediction accuracy at the cost of additional time required for training and prediction. We also built a highly configurable test framework that can be used to efficiently evaluate different prediction approaches and parameterizations.

Convolutional Neural Networks (CNNs) filter the In recent years there has been an effort to reduce the number input data using a series of spatial convolution of parameters in CNNs. Among the first approaches are operators with compact stencils and point-wise the methods of pruning (Hassibi & Stork, 1992; Han et al., non-linearities. Commonly, the convolution operators 2015; Li et al., 2017) and sparsity (Wen et al., 2016; couple features from all channels, which Changpinyo et al., 2017; Han et al., 2016) that have been leads to immense computational cost in the training typically applied to already trained full networks. It has of and prediction with CNNs. To improve been shown that once a network is trained, a large portion the efficiency of CNNs, we introduce lean convolution of its weights can be removed without hampering its operators that reduce the number of parameters efficiency by much.

Behavioral decision theories aim to explain human behavior. Can they help predict it? An open tournament for prediction of human choices in fundamental economic decision tasks is presented. The results suggest that integration of certain behavioral theories as features in machine learning systems provides the best predictions. Surprisingly, the most useful theories for prediction build on basic properties of human and animal learning and are very different from mainstream decision theories that focus on deviations from rational choice. Moreover, we find that theoretical features should be based not only on qualitative behavioral insights (e.g. loss aversion), but also on quantitative behavioral foresights generated by functional descriptive models (e.g. Prospect Theory). Our analysis prescribes a recipe for derivation of explainable, useful predictions of human decisions.