Data Science newsletter – May 24, 2021

What can be done to address these seemingly perpetual issues? Are there specific applications whose use of AI should be limited? And if a fundamental breakthrough in AI leads to an unintended and undesired outcome, should AI-based products be suspended?

Science and engineering have always faced a fundamental challenge that can be distilled into a single question: What is the ethical path for developing and using technology that, although it might enhance the quality of services, might also pose harm to the public? We don’t pretend to know the entirety of that answer. Instead, we offer two frameworks to guide scientists and companies as they pursue the future, including their use of AI. First, we all should remember that basic science is quite different from applied science. And second, we all need to understand the distinction between what science can do and what it should do.

This past academic year, the Year of Engagement taught us the value of staying connected during difficult times and the importance of collaboration. As that campaign culminates, the University of Pittsburgh Office of the Provost has announced that the 2021-22 academic year will be the Year of Data and Society.

The Year Of at Pitt is entering its eighth year of elevating important topics of universal interest across the Pitt community. Past Year Of initiatives have taken deep dives on themes ranging from diversity and sustainability to our health and global impact. Through prolonged focus, the Year Of has challenged how we incorporate creativity into our daily lives and heightened our understanding of the humanities at the University.

Though plans for the Year of Data and Society are still in the early phases of development, Eleanor Mattern, director of the Sara Fine Institute and teaching assistant professor in the School of Computing and Information, has been named campaign chair. In that role, Mattern and a committee of University stakeholders will create opportunities for all members of the Pitt community to engage, explore and experience the Year Of.

In 2018, Cornell researchers built a high-powered detector that, in combination with an algorithm-driven process called ptychography, set a world record by tripling the resolution of a state-of-the-art electron microscope.

As successful as it was, that approach had a weakness. It only worked with ultrathin samples that were a few atoms thick. Anything thicker would cause the electrons to scatter in ways that could not be disentangled.

Now a team, again led by David Muller, the Samuel B. Eckert Professor of Engineering, has bested its own record by a factor of two with an electron microscope pixel array detector (EMPAD) that incorporates even more sophisticated 3D reconstruction algorithms.

Creative Destruction Lab, a nonprofit organization for massively scalable, seed-stage, science- and technology-based companies, will launch its third U.S.-based location, CDL-Seattle, this fall. Based at the UW’s Foster School of Business, CDL-Seattle will be a partnership with Microsoft Corporation, the UW College of Engineering, Paul G. Allen School of Computer Science & Engineering and CoMotion, UW’s collaborative innovation hub. The initial area of focus for CDL-Seattle is computational health.

“The rapid growth of new machine learning applications focused on enhancing human health combined with innovations in sensor technology and other complements has created a flood of new entrepreneurial opportunities that will benefit society,” said Ajay Agrawal, founder, Creative Destruction Lab, and professor at the University of Toronto’s Rotman School of Management. “We’re thrilled to partner with one of the world’s great research institutions, the University of Washington, located in such a vibrant hub of global leaders in technology commercialization — the Seattle region.”

For years, researchers have been trying to find formulae that can accurately predict the future impact of scholarly papers. The latest attempt — a machine-learning algorithm unveiled in Nature Biotechnology on 17 May — has proved controversial.

The tool could help grant funders to identify what research to invest in, inform researchers about promising areas of study and potentially “increase the pace of scientific innovation”, says James Weis, a computational biologist at the Massachusetts Institute of Technology in Cambridge who co-authored the study.

But the paper describing the algorithm has been widely criticized by researchers, some of whom disagree with its suggestion that a mathematical model could be used to help determine which research should receive more funding or resources.

Bad lip-syncing in dubbing and subtitles can put off audiences and hurt box office takings of foreign films.

AI may be about to change all that.

Start-up Flawless AI, co-founded by film director Scott Mann, has a tool that it says can accurately recreate lip sync in dubbing without altering the performance of the actors.

Launched today, a world first project seeks to use artificial intelligence (AI) to predict traffic congestion up to three hours ahead, optimising traffic in large cities and improving road safety as part of the University’s smart cities ecosystem.

University of Melbourne’s Australian Integrated Multimodal EcoSystem (AIMES) brought together PeakHour Urban Technologies, the Victorian Department of Transport, and Telstra to create a large-scale AI application hosted on Amazon Web Services (AWS), which can predict traffic conditions across Melbourne.

Transport engineering expert and AIMES Director Professor Majid Sarvi said the application can also optimise traffic signals for on-road vehicles, freight, and public transport such as buses and trams.

There are few nonprofit startups that can boast Google, Apple and other big tech alumni among their staff. For that matter, there are few startup-style tech nonprofits at all.

But Recidiviz, a tiny startup trying to provide new ways to manage and interpret data for state prison systems, is a nonprofit loaded with staffers who are among the top technical talent in the country. These workers — including CEO Clementine Jacoby, a Google alumna — left their highly-respected tech jobs and took pay cuts to go work for an unknown name because they saw an irresistible prospect: the chance to do work that is both socially important and technically challenging alongside equally expert colleagues.

“All of these people who are like, ‘How do I get into nonprofit tech?’ I don’t often have a great answer. I don’t know. I think there are much fewer opportunities than there is demand for something that is a really meaningful project,” Anna Geiduschek, a Recidiviz software engineer and Dropbox alumna, explained to Protocol. “I would have a hard time working at a nonprofit where the tech caliber wasn’t also quite good. You need both things.”

Grim new birthrate data reflecting conceptions that mostly preceded the pandemic adds to the dire predictions — and worse may be ahead, according to early signs of an even sharper drop since then in people having children.

This means any hope of a rebound is fading.

“The bounce-back isn’t coming, or it’s certainly not going to be coming for a long time,” said Ken Anselment, vice president for enrollment at Lawrence University in Wisconsin.

That’s more bad news for colleges, some of which will continue to struggle to stay afloat; mergers and downsizing of public universities already have been proposed in states including Pennsylvania.

Prices on Amazon have been volatile since the outbreak of COVID-19 in the United States. By the end of March, Keepa’s data revealed, an eight-pack of Barilla spaghetti, whose price normally hovers around $10, shot up to $49.25. The cost of an eight-pack of Skippy Superchunk Peanut Butter nearly quadrupled, from $12.52 to $45.

COVID-19 has made the public more aware of price fluctuations on basics like toilet paper, hand sanitizer, and dried goods, as demand for those products surged beyond expectations.

But limited inventory is not the only reason that the price of pasta (or toilet paper, peanut butter, and rice) is shifting so drastically on Amazon. Experts say the fluctuations on Amazon are often a result of the algorithms sellers use to optimize their sales.

The donation will be split among three different projects.

A $75 million lead grant toward a new $210 million science building located in a prominent spot on Forbes Avenue next to Carnegie Museums.

A new Robotics Innovation Center to be built at Hazelwood Green, fueled by a $45 million grant. The building (which will cost $100 million total) will add 150,000 square feet of space to Carnegie Mellon’s already world-leading robotics research programs.

A $30 million commitment to the Manufacturing Futures Institute, which already operates in Hazelwood Green’s converted steel mill, Mill 19.

NASA is recruiting citizen science volunteers to help astronomers discover exoplanets hidden in observations from one of its space telescopes.

A pair of citizen science projects, Planet Hunters TESS and Planet Patrol, are asking users to help sort through images from NASA’s Transiting Exoplanet Survey Satellite (TESS) and separate out potential exoplanet signals from those of planet impostors. Meanwhile, another project called Exoplanet Watch is tapping amateur astronomers to help follow up on new planet discoveries using backyard telescopes.

Data and data science offer tremendous potential to address some of our most intractable public problems (including the Covid-19 pandemic). At the same time, recent years have shown some of the risks of existing and emerging technologies. An updated framework is required to balance potential and risk, and to ensure that data is used responsibly. Data responsibility is not itself a new concept. However, amid a rapidly changing technology landscape, it has become increasingly clear that the concept may need updating, in order to keep up with new trends such as big data, open data, the Internet of things, and artificial intelligence, and machine learning. This paper seeks to outline 10 approaches and innovations for data responsibility in the 21st century. The 10 emerging concepts we have identified include: End-to-end data responsibility, Decision provenance, Professionalizing data stewardship, From data science to question science, Contextual consent, Responsibility by design, Data asymmetries and data collaboratives, Personally identifiable inference, Group privacy, Data assemblies.

Each of these is described at greater length in the paper, and illustrated with examples from around the world. Put together, they add up to a framework or outline for policy makers, scholars, and activists who seek to harness the potential of data to solve complex social problems and advance the public good. Needless to say, the 10 approaches outlined here represent just a start. We envision this paper more as an exercise in agenda-setting than a comprehensive survey.

I study media representation, marginalized communities and social movements. I have quantified narratives in news coverage of Black civil rights since the 2012 killing of Trayvon Martin, comparing it with coverage of protests for and against former US president Donald Trump, women’s rights, gun control, environmental protection and more. My colleagues and I use computational methods to find linguistic patterns, rhetoric and sentiment in texts, together with human coding for overarching themes including ‘violence’, ‘combativeness’ and ‘racial justice’, as well as for contextual cues, such as the passive voice in headlines, for example “peaceful protesters teargassed”, which neglect to say who took the action.

Linguistic analysis can show what narratives are being presented to and adopted by the public. Such work — examining which groups are privileged at the expense of others — can help many enterprises, including the scientific system, to repair damage from stigmatizing narratives.

Civil-rights protesters are the least likely to have their concerns and demands presented substantively. Less space is given to protesters’ quotes; more space to official sources. Although my work captures amazing individual pieces of journalism that explore themes such as civil rights, protesters’ motivations and communities’ grief, the dominant narrative accentuates trivial, disruptive and combative actions. My early analyses hint that practices improved during the wake-up call that was 2020, but not by much.

It may only be a matter of time until the growing problem of “deep fakes” converges with geographical information science (GIS). Researchers such as Associate Professor of Geography Chengbin Deng are doing what they can to get ahead of the problem.

Deng and four colleagues — Bo Zhao and Yifan Sun at the University of Washington, and Shaozeng Zhang and Chunxue Xu at Oregon State University — co-authored a recent article in Cartography and Geographic Information Science that explores the problem. In “Deep fake geography? When geospatial data encounter Artificial Intelligence,” they explore how false satellite images could potentially be constructed and detected. News of the research has been picked up by countries around the world, including China, Japan, Germany and France.

Microsoft, Amazon and Google are embarked on a frantic race for artificial intelligence (AI) cloud dominance. While in other areas of the cloud space, these technology giants match each other literally feature by feature, machine learning might be the area where they all exhibit highly differentiated capabilities. AWS and Azure are certainly ahead in terms of general customer adoption but machine learning remains one of the strongholds of the Google Cloud offering. This week, at its annual I/O conference, Google unveiled a new series of capabilities that continue to enhance its already robust machine learning offering.

The AI announcements at Google I/O were highly diverse. One of the most interesting was the unveiling of Vertex AI, a new managed cloud service to accelerate the deployment and maintenance of ML models. Vertex AI is a particularly interesting release and it shows significant overlap with Google Cloud ML. Another announcement that captured the headlines was LaMDA, a new conversational model that can produce more natural dialogs by better understanding its context. Google also announced a new generation of Tensor Processing Units (TPUs) chips to power AI workloads as well as new AI-native personalization capabilities to its Firebase mobile backed platform.

BaseTen is the fastest way to build apps powered by machine learning. Deploy models with a few lines of code, serve APIs without infrastructure or framework nightmares, and build stateful, interactive user interfaces to power real, functional applications.