Curated deep tech and AI content that humans at Radical are reading and thinking about. Sign up here to have Radical Reads delivered directly to your inbox every week.

  1. Data and AI: The Real Reason Microsoft Wants to Buy TikTok (Washington Post)

“If AI is the new electricity, the fuel that powers these plants is data,” said Oren Etzioni, chief executive of the Allen Institute for Artificial Intelligence, a Seattle research firm started by Microsoft co-founder Paul Allen.

TikTok has no shortage of data in the form of videos, billions of which are uploaded to the service every year. Those videos offer insight into pop culture. They show people of different ethnicities engaged in a variety of activities. Microsoft can feed all that data into “training models” that help its artificial-intelligence systems learn.

“You don’t really have that fire hose unless you’re plugged in from the inside,” Etzioni said.

And Microsoft’s biggest rivals in developing artificial intelligence systems all have their own video data fire hoses, said Noah Goodman, a professor of natural and artificial intelligence at Stanford University. Google owns YouTube, Amazon owns the Twitch game-streaming service, and Facebook’s massive base of users regularly posts videos.

“For the other big AI players, they have some very large sources of consumer video data,” Goodman said. Microsoft “uniquely doesn’t have access to that among the big AI companies.”

Radical Commentary: We have been asked a number of times lately why Microsoft (‘an enterprise software company!’) would buy TikTok, a consumer video clip service? This article summarizes the explanation we have been giving: Microsoft needs the data to train its AI systems. Video is especially valuable for AI research, training, and applications, including for many use cases that Microsoft can apply across its existing product lines, which include gaming and video conferencing.

Going forward, every acquisition or investment by any large enterprise should be considered through the prism of AI — i.e. does this acquisition advance the buyer’s use of AI? Increasingly the answer will be ‘yes’.

2) AI and AML: The pandemic has changed how criminals hide their cash, and AI tools are trying to sniff it out (MIT Tech Review)

“With many businesses closed, or seeing smaller revenue streams than usual, hiding money in plain sight by mimicking everyday financial activity became harder. ‘The money is still coming in but there’s nowhere to put it,’ says Isabella Chase, who works on financial crime at RUSI, a UK-based defense and security think tank.

The pandemic has forced criminal gangs to come up with new ways to move money around. In turn, this has upped the stakes for anti-money laundering (AML) teams tasked with detecting suspicious financial transactions and following them back to their source.

Key to their strategies are new AI tools. While some larger, older financial institutions have been slower to adapt their rule-based legacy systems, smaller, newer firms are using machine learning to look out for anomalous activity, whatever it might be.”

Radical Commentary: AML tools typically rely on rules-based systems that flag a transaction for verification by a specialist. Increasingly, these systems are both overburdened by the volume of transactions and changes in money laundering behaviour that does not fit clearly within existing AML frameworks.

We expect the future of most anti-money laundering, fraud and other financial security software will be underpinned by machine learning systems able to detect suspicious transactions in real-time. Already there are a number of fast-growing startups in this space, some of which may become massive global players.

3) AI and Human Collaboration: AI is learning when it should and shouldn’t defer to a human (MIT Technology Review)

“Studies show that when people and AI systems work together, they can outperform either one acting alone. Medical diagnostic systems are often checked over by human doctors, and content moderation systems filter what they can before requiring human assistance. But algorithms are rarely designed to optimize for this AI-to-human handover.

…Researchers at MIT’s Computer Science and AI Laboratory (CSAIL) have now developed an AI system to do this kind of optimization based on strengths and weaknesses of the human collaborator.”

Radical Commentary: Optimizing the AI-to-human handover means that the system only defers to its human counterpart if the person could make a better decision. CSAIL’s progress toward accurately predicting when a human could make a better decision contributes to a wider conversation around human oversight required for AI systems and correcting bias.

Although the experiments are relatively simple, such an approach could eventually be applied to complex decisions in areas such as healthcare, where both humans and machines can make errors. The case described in the article, where an AI system helps doctors prescribe the most appropriate antibiotic, highlights the different types of errors humans make compared to AI systems. When facing a trade-off, such as the one between broad-spectrum and specific antibiotics, the AI system could learn to adapt to various doctors with different biases in their prescriptions, and correct for tendencies to over or under prescribe broad-spectrum antibiotics. In this case, optimizing the AI-to-human handover would benefit both the individual patient and the general population.

4) AI and Digital Transformation: Early Lessons from the Coronavirus Crisis (EU Science Hub)

“The pandemic caused something akin to a natural experiment. It has exposed us to unprecedented conditions, forcing us to react in ways unimaginable just six months ago. Four months into this global crisis, we can recognize that COVID has acted as a booster to the adoption of AI but also as an amplifier of potential opportunities and threats.”

Radical Commentary: While COVID-19 has caused massive disruption for many businesses, forward-looking companies have also used it as an opportunity to accelerate the adoption of new technologies, including AI. For publicly traded companies that have historically been forced to manage for quarterly results, the clean sheet provided by the pandemic is an opportunity to invest in the future, and accelerate investment in cutting edge technologies.

5) AI and Education: Improving Online Learning with Artificial Intelligence (EdTech Magazine)

“The academic landscape was already being transformed by economics and technology before the massive disruption of COVID-19. AI can play a key part in this new educational reality, says Goel, a professor of computer science and human-centered computing, and director of Georgia Tech’s Design & Intelligence Laboratory.
…“The old normal is gone forever,” Goel says. “Even when students return to campus, they’ll be going back to more online and blended courses, and we’ll be looking for ways that AI can enhance those classes..”

Radical Commentary: By now it is clear that COVID has accelerated a shift towards digital learning environments. However, the impact of this transformation is particularly apparent in higher education and lifelong learning segments. For years, higher education was increasingly plagued with problems such as increasing costs, reduced access and sub-optimal outcomes. The pandemic has forced a re-evaluation in the minds of learners, especially as alternatives such as upskilling platforms are growing in popularity. Simultaneously, COVID has cracked some of the adoption barriers amongst faculty, enabling new ways of instruction and operations that we expect will proliferate in the months and years ahead.

***

Editor’s Note: We will continue to use this platform to share without commentary articles focused on data and the use of it to illustrate and illuminate racial injustice. Because you cannot fix problems you cannot see or understand.

6) New Report Finds Many U.S. Small & Midsize Cities Face Increased Rent Burden & Income Inequality (NYU Langone Health)

“Small and midsize cities are home to twice as many Americans as large cities and face many of the same health disparities. However, they often do not receive the same attention or resources. A new report by researchers in the Department of Population Health at NYU Langone and NYU Wagner Graduate School of Public Service examining 719 small and midsize cities across the United States paints a picture of uneven economic growth and recovery, growing income inequality and poverty, and excessively high rent burden in nearly every city examined.

To better understand the health and equity trends in small and midsize cities specifically, researchers created a first-of-its-kind, health-focused typology for the 719 U.S. cities with populations of 50,000 to 500,000. This new “City Types” framework groups cities into 10 unique types based on changes in population, household poverty, life expectancy, manufacturing sector employment, income inequality, and other factors that drive health — using data from 2000 to 2017 — and tracked changes over time.

The City Types research builds on the health data for cities available on the City Health Dashboard, a free, online resource created by the NYU Grossman School of Medicine and NYU Wagner Graduate School of Public Service team, and supported by the Robert Wood Johnson Foundation, that provides community-level health, social, and economic data for more than 750 cities across the United States. Careful examination of these new City Types shows that race and poverty affect the opportunities for health for residents in small and midsize cities. And as racial and economic disparities widened over time, so have health disparities.”

— R —