The unprecedented interest, investment, and deployment of machine learning across many aspects of our lives in the past decade has come with a cost. Although there has been some movement towards moderating machine learning where it has been genuinely harmful, it's becoming increasingly clear that existing approaches suffer significant shortcomings. Nevertheless, there still exist new directions that hold potential for meaningfully addressing the harms of machine learning. In particular, new approaches to licensing the code and models that underlie these systems have the potential to create a meaningful impact on how they affect our world. This is Part I of a two-part essay.

The study of the social impact of automated decision making has focused largely on issues of fairness at the point of decision, evaluating the fairness (with respect to a population) of a sequence or pipeline of decisions, or examining the dynamics of a game between the decision-maker and the decision subject. What is missing from this study is an examination of precarity: a term coined by Judith Butler to describe an unstable state of existence in which negative decisions can have ripple effects on one's well-being. Such ripple effects are not captured by changes in income or wealth alone or by one decision alone. To study precarity, we must reorient our frame of reference away from the decision-maker and towards the decision subject; away from aggregates of decisions over a population and towards aggregates of decisions (for an individual) over time. An individual who lives with higher precarity is more affected and less able to recover by the same negative decision than another with low precarity. Thus including only the direct impact of a single decision or a few decisions is insufficient to judge if that system was fair. However, precarity is not an attribute of an individual; it is a result of being subject to greater risks and fewer supports, in addition to starting off at a less secure position. Precarity is impacted by racism, sexism, ableism, heterosexism, and other systems of oppression, and an individual's intersectional identity may put one at greater risk in society, subject to a lower income for the same job, less able to build wealth even at the same income level, and less able to recover from harm.

During the first weeks of February, we asked recipients of our Data and AI Newsletters to participate in a survey on AI adoption in the enterprise. We were interested in answering two questions. First, we wanted to understand how the use of AI grew in the past year. We were also interested in the practice of AI: how developers work, what techniques and tools they use, what their concerns are, and what development practices are in place. The most striking result is the sheer number of respondents. In our 2020 survey, which reached the same audience, we had 1,239 responses. This year, we had a total of 5,154. After eliminating 1,580 respondents who didn't complete the survey, we're left with 3,574 responses--almost three times as many as last year.

The state-of-the-art deep neural networks (DNNs) have significant computational and data management requirements. The size of both training data and models continue to increase. Sparsification and pruning methods are shown to be effective in removing a large fraction of connections in DNNs. The resulting sparse networks present unique challenges to further improve the computational efficiency of training and inference in deep learning. Both the feedforward (inference) and backpropagation steps in stochastic gradient descent (SGD) algorithm for training sparse DNNs involve consecutive sparse matrix-vector multiplications (SpMVs). We first introduce a distributed-memory parallel SpMV-based solution for the SGD algorithm to improve its scalability. The parallelization approach is based on row-wise partitioning of weight matrices that represent neuron connections between consecutive layers. We then propose a novel hypergraph model for partitioning weight matrices to reduce the total communication volume and ensure computational load-balance among processors. Experiments performed on sparse DNNs demonstrate that the proposed solution is highly efficient and scalable. By utilizing the proposed matrix partitioning scheme, the performance of our solution is further improved significantly.

SCI COMMUN### Planetary science The Wright brothers' storied flight at Kitty Hawk had a sequel this week more than 288 million kilometers away: Ingenuity, NASA's $80 million minihelicopter, took a 1-minute test hop on Mars, the first controlled flight of a powered aircraft on another planet. The autonomous 1.8-kilogram machine, the size of a tissue box, spun up its 1.2-meter rotors to more than 2500 revolutions per minute before ascending about 3 meters and hovering in the thin martian air. Ingenuity rotated and took a picture before alighting back on the surface. NASA plans to send Ingenuity, which first landed on Mars on 18 February with the Perseverance rover, on four more flights of increasing height and distance and to use the resulting data to build larger, more ambitious helicopters to explore the Red Planet. 14 of 15 —U.S. states not requiring people to wear masks in public recorded relatively high rates of new COVID-19 cases from May to October 2020. None of eight states with high mask wearing had high rates of infection. ( PLOS ONE ) ### Natural resources Just 19% of Earth's lands are truly wild, with no history of human impact, a new study shows. In other parts of the globe, however, biodiversity hot spots have survived even where humans thrived, thanks in part to millennia of beneficial land management practices by Indigenous people, these researchers conclude. By 10,000 years ago, humans had already spread across three-quarters of the globe, and their controlled burns, small-scale farming, and other practices may have sustained or even improved biodiversity, according to the analysis of past and present land use, published this week in the Proceedings of the National Academy of Sciences . The finding sheds light on a long debate between archaeologists, who cited evidence of this lengthy history, and conservationists, who have insisted that humans did not significantly affect biodiversity until intensive agriculture, urbanization, and deforestation began 200 years ago. Because of the present-day overlap between biodiversity hot spots and lands occupied by Indigenous people, the study bolsters the idea that the growing push to help them regain and retain control over their lands might help protect biodiversity. ### Astronomy The Event Horizon Telescope (EHT) collaboration, which in 2019 produced the first image of a black hole's shadow, this week completed another observing campaign, its first in 3 years. Organizers hope their network of radio telescopes will reveal more of the dark heart of the nearby M87 galaxy as well as the Milky Way's center and the quasar 3C 273. EHT must synchronize 10 observatories across the globe in good weather, so its observing window each year is short. Three observatories joined the network this year (including the Kitt Peak 12-meter telescope in Arizona, below), which will sharpen images. Researchers gathered data for more than seven full nights over 2 weeks this month, and EHT spokesperson Eduardo Ros called the results “excellent.” Now begins a long wait as recorded data are shipped to Boston and Bonn, Germany, for months of processing before an image might be revealed. ### Scientific societies The 90-year-old American Association of Physical Anthropologists (AAPA) has rechristened itself in order to separate today's association from the field's racist and colonial past. At AAPA's virtual annual meeting last week, an overwhelming majority of members voted to delete the word “physical” and become the American Association of Biological Anthropologists. They acknowledged that the old name has roots in the 19th century, when early anthropologists helped create damaging concepts of race by quantifying physical differences among people. The new name conveys that anthropology is now a multidisciplinary biological science that deals with the adaptations, variability, and evolution of humans and their living and fossil relatives, as well as their culture and behavior, according to a statement by the current and past AAPA presidents. “Importantly, the change allows us to reflect deeply on issues of racism and colonialism, which, at times, permeated the field of ‘physical anthropology,’” they wrote. ### Climate science California and its partners announced plans last week to launch two satellites by 2023 to spot plumes of planet-warming carbon dioxide and methane. The $100 million Carbon Mapper project, financed by publisher Michael Bloomberg and other philanthropists, will advance efforts to track concentrated emissions of greenhouse gases that rise from sources such as fossil fuel power plants and leaky pipelines. Previous satellites have lacked the resolution, sensitivity, and focus to collect the data officials need in order to regulate the emissions effectively. The new spacecraft will rely on “hyperspectral” imaging spectrometers that can record more than 400 visible and infrared wavelengths, whose patterns can reveal the abundances of certain gases in the atmosphere below. ### Public health A tiny fraction of the U.S. residents fully vaccinated against COVID-19 by 14 April have become infected, the U.S. Centers for Disease Control and Prevention (CDC) said last week. The agency said it expected some “breakthrough” infections and that the low numbers support the value of the inoculations. CDC said it received 5814 reports of such infections in 75 million people vaccinated in 43 U.S. states and territories. Of the infected people, 65% were female, 45% were 60 or older, and 29% were asymptomatic. Seven percent were hospitalized, and 1% died, some from causes unrelated to COVID-19. CDC cautioned that the data from the states reporting might be incomplete. Public health specialists say the infections were more likely to have resulted from weak immune responses to vaccination than to mutations in the virus that let it evade those defenses. ### COVID-19 Researchers at the University of Oxford will intentionally reinfect people previously infected by the virus that causes COVID-19 to study their immune responses and symptoms. The “human challenge trial,” announced on 19 April, will initially re-expose up to 64 volunteers who previously tested positive for the virus and measure what viral dose triggers new infections. A U.K. government ethics panel approved the study and a similar one led by Imperial College London scientists who are evaluating the performance of COVID-19 vaccines. Such experiments may provide results faster than other trial methods allow. ### Anthropology More than 1300 skulls held in a museum collection that was used to justify racism will now be available for return to communities of the people's descendants, the University of Pennsylvania said last week. Samuel Morton started the collection in the 19th century and used studies of its contents to support the idea of white superiority. Many of the crania belonged to enslaved Africans and Indigenous people. In a statement, Christopher Woods, director of the university's Museum of Archaeology and Anthropology, where the Morton Cranial Collection is held, apologized for the “unethical possession of human remains.” The museum will work to identify descendant communities and accept requests for the return of any crania in the collection. Repatriation of human remains, especially Black and Indigenous ancestors, “is part of a cultural and social reckoning” about how to address anthropology's history of racism, Woods says. ### Scientific meetings A talk last week at the virtual annual meeting of the Society for American Archaeology (SAA) sparked criticism for arguing against a key U.S. law giving Native Americans rights to the human remains and artifacts of their ancestors. Many society members were outraged that SAA gave a platform to what they considered a racist and anti-Indigenous presentation. Some note that this incident comes after a sexual harassment scandal at the organization's 2019 conference. In her talk, SAA member and anthropologist Elizabeth Weiss of San Jose State University said archaeologists “have let creationism into the heart of our discipline” because the law, the Native American Graves Protection and Repatriation Act (NAGPRA), allows Indigenous communities to request repatriation of remains, which they may do partly because of religious beliefs. But archaeologists widely support the law, under which many tribes have collaborated with researchers. In response to the criticism, SAA issued a statement encouraging “the rigorous interrogation of diverse views.” SAA President Deborah Nichols later told Science the organization's board rejects the viewpoint of Weiss and her co-author and supports NAGPRA. ### Policy The relatively modest research investments outlined in Canada's new federal budget could make it difficult for the nation to recruit and retain scientific talent, Canadian science advocates fear. The multiyear spending plan, announced on 19 April, includes CA$2.2 billion in mostly new funding for life sciences, with much of the money aimed at boosting biomedical applications and vaccine development. (Canada will continue to provide other spending for research this year under multiyear budgets approved in 2018 and 2019.) But analysts worry the increases are too modest compared with much larger ones proposed for the United States by President Joe Biden, and that some Canadian scientists will look for work south of the border. Under Canada's budget, three main research councils will share CA$250 million for a new joint biomedical research program, and the Canadian Institutes for Health Research will get an additional CA$250 million to fund clinical trials. Universities and research hospitals will get CA$500 million for infrastructure such as equipment and buildings. Three programs—an existing artificial intelligence program and two new ones in genomics and quantum science—will each receive CA$400 million in new funding. ### Publishing Egypt, Iran, Turkey, and other countries in the Middle East and North Africa have boosted their share of scholarly articles in international journals and citations to those papers during the past 4 decades, the Clarivate analytics firm said this month. From 1981 to 2019, the region quadrupled its share of research articles and reviews to 8%; among regions and large countries, only China grew by more. Fifteen of the region's 19 countries had a citation score in 2019 higher than the world average, when adjusted for differences across disciplines; in 2000, almost all had scores well below average. ### Reckoning with climate blues Sustainability scientist Kimberly Nicholas of Lund University found herself struggling with feelings of grief as research by her and others revealed how much climate change will harm agriculture, ecosystems, and human communities. And she discovered she is not alone. In her new book, Under the Sky We Make: How to Be Human in a Warming World , she offers insight into how people and institutions can respond to those feelings and the climate challenge. (A longer version of this interview is at .) > Q: How does your experience with grief inform your thinking about climate change? > A: Things are changing beyond recognition right now from climate change. To me, grieving is an important part of the process of acknowledging that. It does draw from my experience of losing a dear friend to cancer, who died at 37. It was a kind of wake-up call [that prompted me] to think about my core values and what matters. But it shouldn't take a terminal diagnosis for life on Earth to wake us up to the urgency of working for climate stability. > Q: Students come to you distraught about harm to ecosystems they hope to study. What do you tell them? > A: The main thing is not to shy away from those conversations. It's not really helpful to deny the reality or not equip them with the tools to face that reality. You have to acknowledge that they're running into a house that is on fire. > Q: You argue for a shift from what you call the “exploitation mindset.” What's an example? > A: A big wake-up moment for me came at a climate science conference. Pretty much everyone there, including me, had flown in. The presentations were a litany of depressing things happening because of climate change. I felt like I was at this conference of doctors puffing on cigarettes, but telling our patients to quit smoking! I realized we really have an obligation to model the change that we want to see. So, I have pretty much stopped flying for work. It hasn't meant I can't be a productive researcher.

"For only four hours of simulated data from the telescope, my code could run for five days before outputting the results," Atemkeng says. "I became passionate about the amount of data the SKA would produce in the future – a data stream likely to be in the order of PB/s. At this data scale, even using the most powerful supercomputers, the computation will always remain a major challenge. So I became interested in developing tools for big data, which can be implemented with AI and machine learning algorithms." For Atemkeng, most of Africa's problems – from poverty to food access for its rapidly growing population, education and healthcare access in rural areas – could be solved using machine learning and big data analytics.

Fields such as phylogenetics often work with a sort of abstracted family tree, called a phylogenetic tree, frequently abbreviated here as tree. These trees have different members of a population as their tips, and their branching points describe the relations between the tips and how recently they had a common ancestor. If some of the tips are censored, the tree topology simplifies in a process we refer to as restriction. If one has multiple trees restricted from the same original, uncensored tree, one may wish to reconstruct the original supertree. Suppose instead one has multiple probability distributions of restricted trees, then one may be interested in reconstructing the supertree probability distribution.

In this paper, a novel cognitive architecture for action recognition is developed by applying layers of growing grid neural networks.Using these layers makes the system capable of automatically arranging its representational structure. In addition to the expansion of the neural map during the growth phase, the system is provided with a prior knowledge of the input space, which increases the processing speed of the learning phase. Apart from two layers of growing grid networks the architecture is composed of a preprocessing layer, an ordered vector representation layer and a one-layer supervised neural network. These layers are designed to solve the action recognition problem. The first-layer growing grid receives the input data of human actions and the neural map generates an action pattern vector representing each action sequence by connecting the elicited activation of the trained map. The pattern vectors are then sent to the ordered vector representation layer to build the time-invariant input vectors of key activations for the second-layer growing grid. The second-layer growing grid categorizes the input vectors to the corresponding action clusters/sub-clusters and finally the one-layer supervised neural network labels the shaped clusters with action labels. Three experiments using different datasets of actions show that the system is capable of learning to categorize the actions quickly and efficiently. The performance of the growing grid architecture is com-pared with the results from a system based on Self-Organizing Maps, showing that the growing grid architecture performs significantly superior on the action recognition tasks.

This opinion piece is forthcoming in Il Sole 24 Ore. The Muller-Lyer optical illusion consists of two lines of equal length that differ only in the direction of arrowheads at either end. Yet, to most observers, the line with arrowheads pointing outwards looks longer than the other. If you grew up in and among buildings with straight walls and 90 degree angles, you have learned to perceive lines according to geometric patterns. Your view of the Muller-Lyer lines is biased.

PUNE AI-enabled radiology platform DeepTek is playing an important role in precise diagnosis of diseases like TB and Covid-19. The Pune-based startup has received strategic investment from a clutch of investors so far, and is eying another VC round in next six months. Patil completed his schooling from SSPMS school and engineering from COEP in 1992. He has a Master's degree from IIT-Kharagpur in Industrial Engineering and Operations Research. Amit Kharat, with a DNB and PhD in Radiology, has been engaged in the radiology space for the last 17 years.