Jul 18, 2020

Axios Future

By Bryan Walsh
Bryan Walsh

Welcome to Axios Future, where I'm just waiting for the day when an AI will be able to write the rest of this.

Today's Smart Brevity count: 1,689 words or about 6 minutes.

1 big thing: The future is shrinking

Global population projections to 2100. Credit: The Lancet. Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: a forecasting analysis for the Global Burden of Disease Study

New projections suggest the human population will be smaller and significantly older by the end of the century.

Why it matters: As fertility rates continue to drop around the world, economic and political power among nations will shift rapidly, creating an international landscape radically different than it is in 2020.

Driving the news: A new report in the Lancet by researchers at the Institute for Health Metrics and Evaluation at the University of Washington forecasts the global population will peak in 2064 at around 9.7 billion before declining to 8.8 billion by the end of the century.

  • That slowed growth in the decades ahead followed by outright decline is chiefly caused by drastically falling total fertility rates, from 2.37 women per child globally today to 1.66 in 2100.
  • The fertility rate required to keep population stable is 2.1.
  • Not only will humanity shrink — especially in countries like Japan where fertility has long been below replacement level — it will become much older. By 2100 projections suggest there will be twice as many adults over 80 as there are children under 5.

The big (big) picture: The projected population changes won't play out evenly around the world, which means we could well see major adjustments to the international order.

  • China is set to become the world's biggest economy and is challenging the U.S. for international dominance. But decades of enforced low fertility means that China is set to age and then shrink, potentially setting it on a path of decline shortly after the country reaches the peak of its power.
  • Sub-Saharan Africa is projected to escape population decline, which means that what is currently the poorest part of the world is the one that will have a surplus of an endangered resource: working-age adults.
  • The U.S. is expected to be more resistant to decline than most developed nations, with a population in 2100 projected to rise just slightly higher than current levels. But it will be much older, and those numbers could be skewed if immigration — the main source of growth going forward — is further curtailed.

Yes, but: These projected population declines are in some ways a measure of global success.

  • As countries have grown richer and, especially, as women have become freer and more educated, fertility rates have fallen around the world.
  • A smaller human population should slow the effects of climate change, though it's important to remember that wealth has a much bigger effect on carbon emission than sheer numbers.

The catch: Any forecast that attempts to peer 80 years into the future inevitably rests on assumptions that may be — and almost certainly are — flawed.

  • The UN's latest projections, for instance, are for global population to reach nearly 11 billion by 2100.
  • While economic power has chiefly been driven by large, working-age populations in the past — hence the demographic dividend that has powered growth in parts of Asia and Latin America in recent decades — there's no guarantee that will continue in the future.
  • AI could help economies get far more out of a shrinking population — or possibly, dispense with the need for workers altogether.

The bottom line: By the time my 3-year-old reaches his grandparents' age, the world could be a lonelier place — and a much different one as well.

Go deeper: The aging, childless future

2. Tweeting our way to a nuclear war

Illustration: Sarah Grillo/Axios

A new report outlines how miscalculations and misinformation on Twitter could lead to global war.

Why it matters: Social media platforms like Twitter have vastly accelerated the pace of communication. Without restraints, it's far too easy to imagine how errant tweets could cause international disputes to escalate out of control.

Driving the news: A huge Twitter hack on Wednesday led to the hijacking of authenticated accounts from major figures like Joe Biden and Barack Obama, exposing the vulnerability of a platform that has grown to become a vital platform for real-time information.

What's happening: In a new report coincidentally released the same day as the hack, researchers at King's College London's Centre for Science and Security Studies explored what would happen if those vulnerabilities were exploited on the international level.

  • The "speed, informality and openness" that makes Twitter unique as a communications platform also makes it particularly ripe for misunderstandings.
  • That's a minor issue if the misunderstanding is, say, a feud between two celebrities.
  • It's a much bigger deal if the misunderstanding is between two nuclear-armed nation-states.

Of note: In a 2018 novel, the arms control expert Jeffrey Lewis crafted a fictional scenario where a tweet by President Trump is misinterpreted by Kim Jong-un as a declaration of war, leading the North Korean leader to preemptively launch nuclear missiles at the U.S.

What to watch: The report's authors recommend that the Department of Defense organize an interagency effort to develop best practices for social media use and encourage leaders to refrain from tweeting during crises.

My thought bubble: Axios editor-in-chief Nicholas Johnston has a simple Twitter policy: "Delete your account." We might be safer if more people followed it.

3. Breaking the language barrier in telemedicine

Illustration: Aïda Amer/Axios

As telemedicine use explodes during the pandemic, language interpretation services have lagged behind.

Why it matters: Telemedicine has been a vital lifeline, but if it's going to fully supplement conventional health care, it needs to be able to serve everyone — including people with little or no English skills.

Driving the news: A report this month by FAIR Health found telehealth claims increased by 8,336% nationally between April 2019 and April 2020, rising to 13% of all claims.

  • That increase was almost entirely due to the effects of the pandemic lockdown, which kept patients out of hospitals and doctors' offices.

Yes, but: The nearly 10% of the American population that has limited English proficiency risks being left behind by the telehealth boom, says Kristin Quinlan, the CEO of Certified Languages International. "Too few telehealth platforms are working to build in services for video interpretation."

  • Federal law mandates that any health care provider who received federal funding or reimbursement must provide language access services to patients with limited English proficiency.
  • But that's easier said than done in a country where residents speak hundreds of languages, Quinlan says.

Certified Languages International offers remote interpretation services that can be connected to telehealth calls — including in video, which is important for registering the body language of both providers and patients.

  • Minneapolis-based Allina Health has integrated interpretation into its telemedicine services, notes Frederick Bw’Ombongi, Allina's vice president for access management. "We don't want our most vulnerable patients to be left behind."

The bottom line: It's important to ensure that new technologies are designed to be fair before they become established — not after.

4. Doctors get edge against coronavirus

Illustration: Sarah Grillo/Axios

There's still a lot doctors and scientists don't know about the coronavirus, but they've come a long way since February and March, when they were essentially flying blind, writes Axios' Caitlin Owens.

  • Doctors have learned that flipping patients onto their stomachs instead of their backs can help increase airflow to the lungs.
  • Providers now prefer high-flow oxygen over ventilators, despite the early focus on ventilator supply.

Researchers have also discovered new utility in old drugs:

  • Dexamethasone, a cheap steroid used to treat inflammation, has been found to reduce deaths by one-third among patients on ventilators and one-fifth among those on oxygen.
  • Preliminary data has shown that remdesivir, an antiviral drug, probably doesn't save seriously ill patients' lives but can help others get out of the hospital a few days earlier.
  • Doctors have also learned to put all COVID patients on drugs to prevent blood clots, Armond Esmaili, a hospitalist at the University of California San Francisco Medical Center said.

What we’re watching: These advances in treatment protocols will only go so far, especially if hospitals in states like Florida, Arizona and Texas become too full to put them into practice.

Share this story.

5. Worthy of your time

As risks of space wars grow, policies to curb them lag (Ramin Skibba — Undark)

  • New technology is outpacing regulation as wars on Earth threaten to spread to the stars.

An invisible hand: Patients aren't being told about the AI systems advising their care (Rebecca Robbins and Erin Brodwin — STAT)

  • An in-depth investigation into the fact that around the country, AI systems are involved in making vital decisions about patient care — and patients are in the dark.

Lights out: Live entertainment business struggles to find a way forward amid pandemic (Brent Lang, Jem Aswad, Manori Ravindran — Variety)

  • If you love the theater as much as I do, this piece about its grim future during COVID-19 will be a tough read.

Is it time to kill the penny? (Greg Rosalsky — Planet Money)

  • The answer is yes, please — anything to keep them from showing up in my laundry.
6. 1 AI thing: Automated image generation

OpenAI's iGPT algorithm in action. Credit: OpenAI

An AI algorithm is capable of automatically generating realistic-looking images from bits of pixels.

Why it matters: The achievement is the latest evidence that AI is increasingly able to learn from and copy the real world in ways that may eventually allow algorithms to create fictional images that are indistinguishable from reality.

What's new: In a paper presented at this week's International Conference on Machine Learning, researchers from OpenAI showed they could train the organization's GPT-2 algorithm on images.

  • GPT-2 is best known as a text-generating algorithm, capable of absorbing the structure of language by training on billions of words off the internet and then "writing" passable text from a simple prompt.
  • After training on images, the OpenAI algorithm — now called iGPT — could be fed a visual prompt in the form of half an image and then fill in the rest of it.

How it works: Unlike algorithms that use supervised learning, which requires laborious amounts of labeled data, iGPT and GPT-2 use unsupervised learning on unlabeled data. That means much less human effort.

  • As Karen Hao points out in MIT Tech Review, the fact that both iGPT and GPT-2 use the same algorithm for different purposes is early evidence of more generalizable machine intelligence — which is, after all, OpenAI's aim.

What to watch: A few weeks ago OpenAI released GPT-3, as well as a new API for the tool, which is essentially "a robot that can write anything," as the journalist Alex Hern puts it.

The bottom line: A world where you can ask a robot to write, or create an image of, anything may be just around the corner, which is both exciting and terrifying.

Bryan Walsh

Note: One correction from Wednesday's Future. As one reader pointed out, the Trinity atomic bomb weighed about 5 tons, not 194 tons. Our apologies for the weighty error.