Data Science newsletter – December 18, 2020

GROUP CURATION: N/A

Several years ago, Rachel Hodgdon, an expert in green buildings, was touring a new high school in DeKalb County, Georgia, when she asked teachers how they liked their new building. They loved it. The best part, they told her, was that they no longer went home each afternoon with the “2:30 headache.”

Hodgdon asked them what they meant. “They told me, ‘That’s the term we made up for how sick we feel after a full day at school,’” she says.

At the time, Hodgdon was the director of the Center for Green Schools. As she traveled to meet students and teachers who were moving out of older buildings and into more environmentally friendly ones, she was collecting all sorts of similar stories. Coughs disappeared. Attention improved. Absentee rates dropped.

Warehouses, manufacturing floors, offices, schools — organizations of all kinds have had to change their operations to adapt to life in a pandemic. By now, there is confidence in some ways to help mitigate Covid-19 spread: contact tracing, distancing and quarantining, ventilation, mask wearing. And there is one scientific tool that can play a critical role: testing.

Implementing testing within an organization raises a number of questions. Who should be tested? How often? How do other mitigation efforts impact testing need? How much will it all cost? A new web-based Covid-19 testing impact calculator at WhenToTest.org has been developed by MIT researchers with the Institute for Data, Systems, and Society (IDSS), in collaboration with the Consortia for Improving Medicine with Innovation and Technology (CIMIT), to help organizations around the world answer these questions.

“The calculator allows you to do a cost analysis of different trade-offs and enables rational decisions for deploying testing within an organization,” explains Anette “Peko” Hosoi, a professor of mechanical engineering and IDSS affiliate who co-developed the tool.

Hello and welcome to the last “Eye on A.I.” of 2020! I spent last week immersed in the Neural Information Processing Systems (NeurIPS) conference, the annual gathering of top academic A.I. researchers. It’s always a good spot for taking the pulse of the field. Held completely virtually this year thanks to COVID-19, it attracted more than 20,000 participants. Here were a few of the highlights.

Charles Isbell’s opening keynote was a tour-de-force that made great use of the pre-recorded video format, including some basic special effects edits and cameos by many other leading A.I. researchers. The Georgia Tech professor’s message: it’s past time for A.I. research to grow up and become more concerned about the real-world consequences of its work. Machine learning researchers should stop ducking responsibility by claiming such considerations belong to other fields—data science or anthropology or political science.

Humans are emotional beings, with feelings that show in our behaviours and facial expressions. But whether these mean the same thing in different cultures has been hotly debated.

In what researchers say is the first worldwide analysis in naturalistic settings, a new study published in the journal Nature has found that different social contexts, such as weddings, funerals, humour, art and sports, do indeed elicit universal facial expressions.

“We found that rich nuances in facial behaviour – including subtle expressions we associate with ‘awe’, ‘interest’, ‘triumph’ – are used in similar social situations around the world,” says lead author Alan Cowen from the University of California Berkeley, US.

For scientists, 5 January was a turning point in the fight against the coronavirus. That day, a team led by Prof Yong-Zhen Zhang at Fudan University in Shanghai sequenced the genetic code of the virus behind Wuhan’s month-long pneumonia outbreak. The process took about 40 hours. Having analysed the code, Zhang reported back to the Ministry of Health. The pathogen was a novel coronavirus similar to Sars, the deadly virus that sparked an epidemic in 2003. People should take precautions, he warned.

The Chinese government had imposed an embargo on information about the outbreak and Zhang and his co-workers were under pressure not to publish the code. The blackout couldn’t hold. On 8 January, news broke about the nature of the pathogen and was confirmed a day later by Chinese authorities. To sit on the code now seemed ridiculous.

Eddie Holmes, an evolutionary biologist at the University of Sydney, and a collaborator of Zhang’s, called him to push for publication. Zhang was buckling up on a flight bound for Beijing. As the plane left the runway, they two agreed to break the gagging order. On 11 January Australia time, the day China announced its first official death from the infection, Holmes published the sequence on a website called virological.org. It was a crucial act for researchers around the world. Holmes calls it “ground zero for the scientific fight back against the disease”.

When Matt Noblett, a partner in the architecture firm Behnisch Architekten, started working on designs for a biomedically focused research facility in Allston in 2007, the School of Engineering and Applied Sciences—many of whose faculty members are now moving into an entirely reimagined, spectacular new science and engineering complex on the site—did not even exist. (SEAS, part of the Faculty of Arts and Sciences, was established as a school later that year.) After the financial crisis of 2008-2009 struck, and the original four-building plan was scrapped, Noblett and colleagues again worked on the site, this time waterproofing and sealing the unbuilt structure’s five-acre basement to preserve it for an uncertain future use. When called back a third time in 2014, the group began work on an entirely new concept: a new home for half the departments within a rapidly growing SEAS that would include laboratories and shared core facilities, classrooms, makerspaces, and dedicated areas for undergraduates working on projects such as solar-powered cars. This third time—though bittersweet due to the pandemic’s enforced dampening of what should have been a celebratory opening—is indeed the charm. The new 544,000-gross-square-foot science and engineering complex (SEC), at 150 Western Avenue facing the Business School campus, is, in the words of Charles River professor of engineering and applied sciences Robert (Rob) Wood, a “wonderful jewel of a space.”

Our new study uses real-time data supplied by Indeed, one of the major providers of worldwide employment-related search engines for job listings. This gives us one-of-a-kind insight into the behavior of virtually all employers that post online job advertisements. The main advantage of Indeed’s online job posting data is that the information is close to real-time data and offers complete coverage of online job postings, whereas government survey data are limited to the employers surveyed. This real-time view of labor demand provides analytical backing for something that has become increasingly evident as the year has progressed: demand for jobs for women has fallen disproportionately more than for men, and low-skilled workers are likely to fall further behind.

The National Research Council of Canada (NRC) is pleased to announce a partnership with Simon Fraser University (SFU) to establish a West Coast collaboration to increase capacity in artificial intelligence (AI) and data science research and training in Canada.

This new partnership further strengthens SFU’s AI and data capabilities and builds upon a history of collaboration with the NRC. With a shared commitment to data-driven research, it is positioned to bring together researchers, governments, industries and communities to deliver innovations across many sectors in Canada.

The twelve articles in this special theme fall into three categories. The first group comprises four articles that provide context and refection on the NASEM report and reproducibility and replicability. … The next set of articles began life as commissioned papers by the NASEM Reproducibility and Replicability in Science committee. Each of these five articles informed the committee’s deliberations and was subsequently reviewed and revised (some rather substantially) for publication in HDSR. … The final group of articles are research articles that extend or analyze the report recommendations or conclusions.

On Monday, December 14, I caught up with Gebru via Zoom. She recounted what happened during her time at Google, reflected on what it meant for the field and AI ethics research, and gave parting words of advice to those who want to keep holding tech companies accountable. You can also listen to a special episode of our podcast, In Machines We Trust, for highlights from the interview. (Google declined a request for comment on the contents of this interview.)

Amazon SVP of Devices & Services David Limp joined us at TC Sessions: Space today, and he shared some new details about the company’s Project Kuiper broadband satellite constellation. Limp shared more details on the technical design challenges that the Kuiper team solved with its revolutionary customer terminal, but he also shared more info on the company’s plans around launching its constellation, which will number 3,236 per the current plan approved by the FCC.

“We’re launch agnostic” Limp said. “If you know somebody who has a rocket out there, give us a call. “One of the reasons we thought the time was right to do a constellation now is because of some of the dynamics happening in the launch industry. Every day, we see a new demonstration of reusability, every day we see new demonstrations of breakthroughs in better engines, whether that’s Raptor [SpaceX’s engine] or BE-4 [Blue Origin’s].”

Part of the FCC’s approval for Amazon’s constellation requires it to send up around half of its planned total constellation within the next six years, which is a significant volume and will require an aggressive launch pace to achieve.

A recent study has identified four factors that predict user satisfaction with customer service chatbots. The study also found that a positive chatbot experience was associated with customer loyalty, highlighting the potential importance of the findings to corporate brands.

“Chatbots that use artificial intelligence to address customer needs are already in widespread use, and are expected to become even more common over the next few years,” says Yang Cheng, corresponding author of the study and an assistant professor of communication at North Carolina State University. “We wanted to know how these chatbots are affecting the user experience and how they affect the way consumers feel about brands.”

For this study, researchers conducted a survey of 1,064 consumers in the United States who had used a chatbot from at least one of the 30 U.S. brands with the most highly regarded chatbot services. These brands represent industries ranging from fashion to food to software.

Federal regulators are ordering Facebook, Twitter, Amazon, TikTok’s parent and five other social media companies to provide detailed information on how they collect and use consumers’ personal data and how their practices affect children and teens.

The Federal Trade Commission’s action announced Monday goes to the heart of the tech industry’s lucrative business model: harvesting data from platform users and making it available to online advertisers so they can pinpoint specific consumers to target.

The agency plans to use the information, due in 45 days, for a comprehensive study.

With average temperatures climbing and severe weather events happening more frequently, the Fed is increasing its consideration of how environmental shifts could create risk for the financial sector.

In its most recent financial stability report, the Fed included a section dedicated to the issue of climate change.

The year 2020 will enter the history books as the year in which a new deadly coronavirus brought the world to a halt. Pharmaceutical companies jumped to the rescue with major investments in vaccine research and development. Last month, one pharmaceutical company after the other started releasing insights about the efficacy of their candidate vaccines. While these announcements have major implications for the world’s economy in 2021, they also provide valuable lessons for managers who want to use data to make better decisions.

Lesson 1: Big data is often smaller than it appears.