A zettabyte is a trillion gigabytes. That's a lot--but, according to one estimate, humanity will produce a hundred and eighty zettabytes of digital data this year. It all adds up: PowerPoints and selfies; video captured by cameras; electronic health records; data retrieved from smart devices or collected by telescopes and particle accelerators; backups, and backups of the backups. Where should it all go, and how much of it should be kept, and for how long? These questions vex the computer scientists who manage the world's storage. For them, the cloud isn't nebulous but a physical system that must be built, paid for, and maintained.

AI text-to-image software like Midjourney, DALL-E and Stable Diffusion has the potential to change the way that architects approach the creation and concept stages of designing buildings and products, experts say. In the past year, numerous technology companies have released software that uses AI systems called neural networks to convert users' text inputs into AI-generated images. The images created by these bots – including eerily real-looking imaginary buildings – have become an internet sensation and led to discussions about how they could impact the future of design and architecture. Bill Cusick is the creative director at one such company, Stability AI – which has released text-to-image software called Stable Diffusion and DreamStudio – and also has experience working with the software of the popular visualisation company Midjourney. He believes that the software is "the foundation for the future of creativity".

Manas Bhatia has a bold vision of the future -- one where residential skyscrapers covered in trees, plants and algae act as "air purification towers." In a series of detailed images, the New Delhi-based architect and computational designer has brought the idea to life. His imagined buildings are depicted rising high above a futuristic metropolis, their curved forms inspired by shapes found in nature. But the pictures were not entirely of his own imagination. The architect's conceptual towers were created using AI imaging software.

> Science's COVID-19 reporting is supported by the Pulitzer Center and the Heising-Simons Foundation California was a COVID-19 success story—until suddenly it wasn't. Early in the pandemic, the state seemed to have the new coronavirus under control, but it has begun to ride a wave there, with records set in daily cases several times this month, and deaths on the rise. California officials whose COVID-19 responses were once hailed as enlightened are now receiving criticism—and some of the sharpest is coming from scientists seeking to help guide the state's fight against the virus. Since April, epidemiologists from Stanford University and several University of California (UC) campuses have sought detailed COVID-19 case and contact-tracing data from state and county health authorities for research they hope will point to more effective approaches to slowing the pandemic. “It's a basic mantra of epidemiology and public health: Follow the data” to learn where and how the disease spreads, says Rajiv Bhatia, a physician and epidemiologist who teaches at Stanford and is among those seeking the California data. But the agencies have refused requests filed from April through late June, Science has learned. They cited multiple reasons including workload constraints and privacy concerns—even though records can be deidentified, and federal health privacy rules have been relaxed for research during the pandemic. As a result, Bhatia says, “In 4 months of the epidemic, collecting millions of records, no one in California or at the CDC [U.S. Centers for Disease Control and Prevention] has done the basic epidemiology.” Other states also fail to share highly specific information for their COVID-19 cases, which some scientists warn is hampering efforts to identify targeted measures that could stem the spread of SARS-CoV-2 without full-scale lockdowns. Bhatia and epidemiologists across the country are especially aggrieved after recent news reports revealed states are feeding the same data they desire to a federal contractor, Palantir Technologies, that has drawn criticism for data work supporting Immigration and Customs Enforcement deportations. For a data platform dubbed HHS Protect, Palantir is aggregating information on the spread of the new coronavirus on behalf of the U.S. Department of Health and Human Services (HHS), drawing on more than 225 data sets, including demographic statistics, community-based tests, and a wide range of state-provided data. (This week, sparking concern among public health experts, epidemiologists, and others, HHS also instructed hospitals to provide data on COVID-19 cases and patient information directly to the Palantir system—largely via a second contractor—rather than to CDC as they have for decades.) Aggregated COVID-19 case and death data by county, and often by age and race, are publicly available in much of the country. But few locales link those cases and deaths to other information typically collected on the individuals, such as ZIP codes, occupations, living conditions, and known contacts with others ill with COVID-19. A survey of public data dashboards for all 50 states, Washington, D.C., and Puerto Rico by Prevent Epidemics, a group led by former CDC Director Tom Frieden, found that just 2% of data for 15 key COVID-19 indicators were fully reported. Only 40% of the data were partially reported—with glaring deficiencies for testing and contact tracing. Bhatia and colleagues say that detailed COVID-19 case data could be mined to find factors most responsible for the “biggest bundles of hospitalizations and deaths.” He hypothesizes the data would, for example, confirm that even as commerce opens up, hospitalizations and deaths mostly emerge from familiar flashpoints. He cites care facilities for the elderly and large households that include infected essential workers who are asymptomatic or have mild symptoms; they may then pass the disease to relatives who have risk factors making them more vulnerable to severe illness. “We think you can be more strategic on your interventions if you know where exposures actually occur,” says Jeffrey Klausner, a physician and epidemiologist at UC Los Angeles, who is also seeking his state's data. For example, case data might confirm patchy evidence that indoor dining is risky, but parks and beaches are generally safe. If so, reopening outdoor settings with reasonable precautions might boost the economy and allay fears that severe risk of infection is ubiquitous. As the pandemic evolves, regular reassessment of granular data on cases is vital, says Natalie Dean, a University of Florida (UF) biostatistician. “We have this whole new world now, where we are opening things back up. We have this shifting set of environments—indoor dining, bars, open retail buildings, offices, gyms. When we think of what are pressure points, there's a lot we just don't know yet. … We have to have ‘a learning architecture’ in place where there's always some level of reflection.” In the absence of clear, localized data from public authorities, some clinics in California have done their own research. After conducting thousands of COVID-19 tests in Oakland, “We have been able to pinpoint where some of the outbreaks are, both geographically and in terms of setting,” leading to highly targeted health education and testing outreach, says Noha Aboelata, a physician who heads the city's Roots Community Health Center, which primarily serves people of color in underserved communities. Without neighborhood-level intelligence for public health outreach, you get “a one-size-fits-all solution that might exacerbate the problem,” she says. “Withholding the information is going to lead to deaths.” In response to Science 's questions, the California Department of Public Health wrote that even deidentified data “can be used alone or in combination with publicly available information to identify an individual.” Caitlin Rivers, an epidemiologist at Johns Hopkins University's Center for Health Security, calls reidentification a valid concern, but argues it would happen so rarely that the risk shouldn't justify blanket denials of data requests during the pandemic. “There's a lot of space in the middle that we haven't really explored,” she adds. For example, to obviate some privacy concerns, Bhatia's group requested case reports giving 10-year age ranges rather than specific ages, the week of COVID-19 onset rather than a specific date, and an occupational group rather than specific occupation. To show the value of richer data, Bhatia turned to Florida, which offers fairly detailed information on each of the more than 316,000 COVID-19 cases recorded there so far. The data set enabled him to graph, week by week, infections by age and whether the source of transmission was known. He found that early in the pandemic, the source was known for 80% of children, and 50% to 60% of adults. As Florida relaxed restrictions on businesses and other aspects of life, known sources of transmission remained at similar levels, even though casual contact with strangers was apparently increasing. Because some of the unknown sources of transmission were certainly asymptomatic or mildly symptomatic family or friends, such a finding suggests crowded beaches are playing a smaller role in Florida's surge in infections than, say, increased numbers of large family gatherings at home or repopulated offices. “If people know that 50% or 60% of infections are resulting from people they know, including family, friends, and co-workers, they may better interpret risk,” Bhatia says. Even Florida's data exclude key details that some researchers view as essential to map and respond to the pandemic most effectively—including ZIP codes; more complete racial designations; and specifics on cases in long-term care facilities, jails, and prisons. That hampers targeted responses, says Thomas Hladish, an infectious disease researcher at UF who consulted extensively with state officials about COVID-19 data from March until this month. “A lot of the inconsistencies that you see are reasonably explained by well-intentioned people who are scrambling to reinvent [data fields and formats] on the fly without the appropriate technical background.” The Miami Herald also recently reported that municipal officials have not been able to get the state to provide case details they need to attack local outbreaks. The Florida Department of Health did not respond to Science 's requests for comment. Epidemiologists praise more forthcoming agencies. The New York City Department of Health and Mental Hygiene posts unusually complete, continually updated data sets on COVID-19—showing detailed information on tests, cases, and deaths for 177 discrete neighborhoods—and uses them to map hot spots. It offers probable and confirmed deaths by age, race or ethnicity, underlying conditions, and other factors. One clear finding: Lower income areas, with a higher concentration of large households, suffered from COVID-19 at many times the rate of most wealthy areas. The city's health commissioner, Oxiris Barbot, says the system was crucial in decreasing cases by about 94% and deaths by about 98% since they peaked in April. “The transparency in data helped to paint a picture of how acute a situation we were in and the degree to which we needed New Yorkers to comply with what we were asking them to do,” she says. “It helped highlight as early as possible the ways in which the virus was ravaging Black and brown communities.” And the granular data allowed a calibrated response—including offers of hotel rooms to help people living in crowded conditions isolate when diagnosed with COVID-19. “Had it not been for that data analysis we would have been much slower in the response, and … many more lives would have been lost,” Barbot says. “These are the right type of efforts using the right type of data,” Bhatia says. Figuring out how to stop the pandemic is “the biggest and most impactful policy decision we've seen in our lifetimes,” he adds. But in California and elsewhere, “We're trying to predict the future without analyzing the data that's in front of us. That's a failure.”

On Saturday, the Ministry of Electronics and Information Technology (MeitY) entered into a collaboration with Google to launch "Build for Digital India," a programme that will provide learners in engineering with a platform for market-ready development technology-based solutions that address key social problems. The programme will invite ideas from students to focus on areas of healthcare, agriculture, education, smart cities and infrastructure, women safety, smart mobility and transportation, environment, accessibility and disability and digital literacy. As part of the programme, engineering students across the country will be invited to apply and join in a learning journey that will help them transform their bright ideas into real-world solutions. Applicants will take part in online and offline learning opportunities on key technologies such as Machine Learning, Cloud and Android. These will be offered through Google's Developer Student Club network and other Google Developer networks.

The goal is to capture the sensitivity of inferential proc edures to the presence of outliers in the data and misspecifications in the modelling a ssumptions, and to mitigate overly large sensitivity. The Bayesian approach has been fo cused on capturing possible anomalies in the observed data via the model and in choosing p riors that have minimal effect on inferences. The frequentist approach, on the other hand, has focused on the development of estimators that identify and guard against o utliers in the data. We refer the reader to [ Hub11, Chap 15] for a comprehensive discussion.

MIT engineers have designed tiny robots that can help drug-delivery nanoparticles push their way out of the bloodstream and into a tumor or another disease site. Like crafts in "Fantastic Voyage" -- a 1960s science fiction film in which a submarine crew shrinks in size and roams a body to repair damaged cells -- the robots swim through the bloodstream, creating a current that drags nanoparticles along with them. The magnetic microrobots, inspired by bacterial propulsion, could help to overcome one of the biggest obstacles to delivering drugs with nanoparticles: getting the particles to exit blood vessels and accumulate in the right place. "When you put nanomaterials in the bloodstream and target them to diseased tissue, the biggest barrier to that kind of payload getting into the tissue is the lining of the blood vessel," says Sangeeta Bhatia, the John and Dorothy Wilson Professor of Health Sciences and Technology and Electrical Engineering and Computer Science, a member of MIT's Koch Institute for Integrative Cancer Research and its Institute for Medical Engineering and Science, and the senior author of the study. "Our idea was to see if you can use magnetism to create fluid forces that push nanoparticles into the tissue," adds Simone Schuerle, a former MIT postdoc and lead author of the paper, which appears in the April 26 issue of Science Advances.

Artificial intelligence (AI) is everywhere. As a rapidly developing tech that is integrating itself into every aspect of our lives, it exists in the virtual assistants living inside our smart homes through the Internet of Things (IoT) and by extension through our smartphones. As consumers, it drives our interactions with brands. But when it comes to business, what is encouraging this widespread adoption? With AI technology, it's easier than ever to gather data and draw actionable conclusions that can increase revenue, drive business growth goals and establish your business as a leader in the industry.

In this paper, we study the problems of principal Generalized Eigenvector computation and Canonical Correlation Analysis in the stochastic setting. We propose a simple and efficient algorithm, Gen-Oja, for these problems. We prove the global convergence of our algorithm, borrowing ideas from the theory of fast-mixing Markov chains and two-time-scale stochastic approximation, showing that it achieves the optimal rate of convergence. In the process, we develop tools for understanding stochastic processes with Markovian noise which might be of independent interest.

Indian e-commerce companies are still novices when it comes to Artificial Intelligence (AI) and mining big data. Nevertheless, they are still betting big on AI, which they think could be the magic bullet that will help them offer tailored shopping experiences. In 2017, homegrown e-commerce firm Flipkart announced the launch of an initiative called AI for India, wherein the company would develop solutions for issues like deciphering complex addresses and catching address fraud. Flipkart-owned fashion brand Myntra runs two AI-powered brands, Moda Rapido, and Here and Now. Delhi-based Paytm's homepage is personalised and reordered differently for each of its 225 million users, and the platform makes 20,000 recommendations per second--each of them in under 20 milliseconds.