1 big thing: Rising AI superpowers
In the 19th century, the Industrial Revolution cemented Great Britain's claims to global superiority, and later catapulted the United States into dominance. But they were not alone. Japan and Germany also arose as major industrial, military and political powers.
What's going on: In a transition that could be as momentous, the U.S. and China today are racing to master artificial intelligence. But it's no longer clear, as had seemed the case, that one or the other will hold the field to itself. According to a new study, the winner may be forced to share geopolitical sway with smaller nations like Israel, Russia or Singapore.
The study, released by the Center for a New American Security, suggests that the U.S. government has been flat-footed in the race, and reprises a refrain that Beijing is richly supporting its own private players.
- "The U.S. hasn't really been paying attention to sustainable sources of superiority in AI, so China could develop a first-mover advantage that would be sustainable," Gregory Allen, a senior fellow at CNAS and co-author of the report, tells Axios.
- And if AI turns out to be easy to copy and replicate, the U.S. needs to be prepared to adapt to a diffused tech ecosystem in which AI is in the hands of smaller countries.
- The report specifically mentions India, Israel, Russia, Singapore and South Korea. To the degree AI includes more players, it increases the "risk that countries may put aside the safety and reliability concerns" that experts have expressed given AI's capacity for social and political disruption.
The U.S. is ready for neither potential outcome, the report says.
One question is why experts think that AI will confer unusual geopolitical leverage. The report says:
- AI is akin to electricity and the combustion engine, general applications that triggered a shower of cross-cutting technologies, massive surges in productivity, and negative consequences like social and political instability.
- That makes AI unlike nuclear weapons or the battleship, which were momentous but less dispersed in their applicability and impact.
- "We have been waiting for the United States government to make a response commensurate to the scale of what we're observing," Allen said.
2. How Uber, Lyft make traffic worse
In the 1930s, New York building commissioner Robert Moses built one highway and bridge after another, with the aim of relieving congestion in America's biggest city. But each time, the result was the same: worse traffic.
What's going on: Eight decades later, transportation experts are observing a similar phenomenon with the world's newest urban innovation: ride-hailing services.
- Axios' Henrietta Reily and I write today: According to a major new study, Uber, Lyft and their smaller rivals are clogging major U.S. cities, not relieving congestion, and even more traffic may be on the way when self-driving cars are commonplace.
Why it matters: A major promise of the self-driving, ride-hailing future has been cleaner, more walkable, and people-friendly cities, with much more efficient, individual transportation. But if the study — like others before it — is accurate, we are instead heading toward a bigger problem.
Bruce Schaller, a former New York deputy commissioner of transportation and author of the report, tells Axios that when people use a ride-hailing company, they are opting to do so rather than take public transportation, walk or bike. They generally are not choosing between hailing and driving themselves.
- U.S. ridership is surging, he said — up 37% last year, to 2.6 billion ride-hailing passengers, from 2016.
- And hailing added 5.7 billion miles of driving a year to the nine cities in the study — Boston, Chicago, Los Angeles, Miami, New York, Philadelphia, San Francisco, Seattle and Washington.
- Uber and other ride-hailing services may not have exacerbated traffic initially. "But now they are clearly a source of congestion, and to deal with congestion you have to deal with them," he said.
Schaller's report aligns with an October study released by UC Davis. It found that, in U.S. cities, 49% to 61% of ride-hailing trips would have not been made at all, or by walking, biking, or public transit.
- Regina Clewlow, a transportation research scientist and an author of the UC Davis study, told Axios that no one expected such consumer demand for the rides.
- "Cities were blindsided by the dramatic growth of ride-sharing companies," she said.
3. AI blunders
AI experts — concerned about reported blunders with high-stakes systems from Amazon and IBM — are urging more oversight, testing, and perhaps a fundamental rethinking of the underlying technology, Kaveh Waddell writes.
- Amazon’s face-recognition platform, Rekognition, matched 28 members of Congress with mugshots when it was put through testing by the ACLU, which announced the results Thursday. The misidentified faces disproportionately belonged to people of color.
- IBM’s Watson gave doctors "unsafe and incorrect" recommendations for cancer treatments, Stat News reported last week, quoting internal IBM documents. The finding blamed both IBM engineers and the doctors who were feeding in training data.
In an earlier case, a self-driving Uber killed a pedestrian in Arizona in March.
- Responding on its blog, Amazon said the ACLU didn’t test Rekognition with the correct settings, and that its system is meant to help humans make big decisions, not final determinations on its own. And IBM told Stat News that Watson Health has since improved.
But there are reasons experts are worried: Wall Street, the military, and other sectors expect AI to make increasingly weighty decisions in the future, with less and less human involvement. And if the systems behave inaccurately or display biases, the consequences outside the lab could cause harm to real people.
- For skeptics of deep learning, the leading machine-learning method that powers most commercial AI, these shortcomings belie greater problems ahead.
- "We shouldn’t mistake pattern recognition for genuine intelligence," Gary Marcus, an NYU professor, tells Axios in an email. "And we shouldn’t be surprised when narrow, shallow intelligence (which is all we have, so far) lets us down."
- Garrett Kenyon, a scientist at Los Alamos National Laboratory, said deep learning can’t grasp abstract concepts, or even reliably count or compare objects.
Jack Clark, strategy and communications director at OpenAI, says these cases are not marks against deep learning as a technology — just indications that the method was poorly implemented in this case. Speaking to Axios, he urged that AI systems be "vigorously and transparently tested in the wild" before they’re put to work in the real world.
4. Worthy of your time
- When the trade gates opened to China (Bob Davis - WSJ)
- 127 degrees in the shade (Andrew Freedman - Axios)
- A podcast interview with Jaron Lanier (Kara Swisher - Recode)
- Imagining a full-throated gig economy (The Economist)
- Enroll, photograph, replicate (Cynthia McFadden, Aliza Nadi, Courtney McGee - NBC)
- Seoul's new short, 52-hour work week (Su-Hyun Lee, Tiffany May - NYT)
5. 1 dog thing: A drug-sniffing Colombian media star
Among Colombia's pantheon of media stars is Sombra, a drug-sniffing German shepherd with a $7,000 bounty on her head — dead or alive — from the irate Gulf Clan cocaine cartel, reports Manuel Rueda of The Associated Press.
What's going on: Over her career, Sombra, working with the Colombian police, has pointed the way to more than two tons of cocaine.
- Sombra is skilled at airports, where she has contributed to a reported 245 drug-related arrests.
- In one case, she found tons of cocaine hidden among bananas on their way to Europe.
- In another, she found 170 pounds of the drug hidden within machinery.
That record has naturally ignited the ire of the Gulf Clan, so in January, police shifted her duties. Now Sombra sniffs out illicit cargo at El Dorado Airport in Bogota. When she goes off duty, she often has two police guards.
- Jose Rojas, her handler, tells the AP: "Her sense of smell is far beyond that of other dogs."