Axios Future

A robotic hand with the palm facing upward.

October 17, 2018

Welcome back to Future. Thanks for subscribing.

Consider inviting your friends and colleagues to sign up. And if you have any tips or thoughts on what we can do better, just hit reply to this email or shoot me a message at [email protected]. Okay, let's start with ...

1 big thing: Small, narrow — and revolutionary

Illustration of robot arms holding a scalpel, a wrench and a light bulb.

Today’s best AI algorithms, which can perform single tricks, are years away from "general intelligence." Illustration: Aïda Amer/Axios

Talk of artificial intelligence goes almost inexorably to the very large. Companies, it is said, must embrace big data, along with future human-like AI, or be lost to history. But some tech leaders are going the other way, urging businesses to start thinking small.

What's happening: This growing mantra speaks of the upside of narrow AI ambitions and small data. But by small bore, they don't mean small results. AI and data minimalism, they say, could be what revolutionizes business, industry, war and more, writes Axios' Kaveh Waddell.

  • The dream of most AI researchers is "general intelligence" — the broad human capacity to cook, do a crossword, work at a computer and carry on conversations with friends, all seamlessly, one after the other.
  • But that's a distant dream — researchers are at least decades away from general intelligence.
  • And that's not even contemplating super-human intelligence, the holy grail.

Instead, today’s best AI algorithms are one-trick ponies that have each been taught an extremely useful, single trick, Andrew Ng, an AI pioneer at Stanford University, tells Axios.

  • Examples are algorithms that drive cars, read chest X-rays and translate languages.
  • None can do any other thing — only its one task.

But that's not something to sneer at: These ponies have already created more than $1 trillion of business value, according to Gartner — with many trillions more up for grabs for those who can figure out how to apply the young technology.

  • We’ve written of some of the problems that confound the best AI systems, but in many areas, AI can outperform humans.
  • "Almost anything you can do with less than a second of mental thought, we can probably now automate" with AI, says Ng, who co-founded Google Brain and Baidu's AI program.

The big picture: Driving the optimism is deep learning, the technique that has in recent years been the jet fuel accelerating AI applications, and it shows few signs of slowing.

  • "We're in year two or year three of a good, 40-year run," says Frank Chen, a partner at Andreessen Horowitz, a prominent Silicon Valley VC firm.
  • Even if, by some dictatorial decree, all deep learning research stopped today, developers would be writing new software using today’s technology for decades, says Chen. "We have a long way to go just harnessing the existing techniques."

Deep learning has its own drawbacks. Among the most constraining is its never-ending hunger for more and more data.

But some problems don’t come with a lot of data. Ng offers some examples:

  • A factory that wants to nab defective parts before they roll off the manufacturing line may have just a handful of examples with which to train an algorithm.
  • A hospital that wants to detect diseases from medical imaging may only have a small number of scans on hand for a particularly rare condition.

The bottom line: “I think big data is overhyped,” says Ng.

  • A lot of problems will be solved using small data. The only problem? It takes “a much more skilled team to do things with small data,” he said.
  • One way to overcome the challenge is to pair deep learning, which uses correlations to make predictions, with explicit rules that can clear up ambiguity, writes Virginia Dignum, a computer science professor at Delft University of Technology.

Go deeper:

2. Consider the parking lot

Photo: Jan Woitas/picture alliance/Getty

America is overparked. In Los Angeles, for example, there are 9 parking spaces for every car. Nationally, 250 million adults have access to more than 700 million parking spaces. That adds up: The U.S. dedicates an area the size of Connecticut to parking, Jody Kelman, director of the Self-Driving Platform team at Lyft, writes for Axios Expert Voices.

What's happening: When proponents of self-driving cars speak, they talk of getting rid of a lot of those spaces and turning them into something else, such as dedicated bike and scooter lanes, on-street parklets and housing.

Fun fact: They also speak of saved time. In Westwood Village, a shopping strip in Los Angeles, consumers spend approximately 95,000 hours each year circling for parking.

  • That’s 11 years of wasted time, in just one small stretch of roadway.

This does not mean that self-driving cars won't park at all. But when they need to (at low-demand hours, for instance), they can do so more precisely than a human.

  • Parking lots currently budget around 325 square feet per car, but Audi estimates that self-driving cars will need a space 30% smaller.

What to watch: As shared and autonomous vehicles become a larger part of the national fleet, the footprint of cities will change with them.

3. MIT investing $1 billion into AI research

A domed building on MIT’s campus

Photo: William B. Plowman/Getty Images

In its first fundamental restructuring in nearly 70 years, MIT has announced that it’s pumping $1 billion into creating a new college focused on computing and artificial intelligence.

Why it matters: MIT is planning what it says will be the single largest investment in computing and AI by an American university at a time when the U.S. and China are competing to produce — and retain — top AI talent in order to reap the technology's economic and geopolitical gains.

Details: The new college, slated to open next year, will eventually be staffed with 50 new faculty members, half of whom will be appointed to both the college and another MIT department. Additional faculty will come from elsewhere in the university.

  • The hope is to connect parts of the university that have been siloed from MIT's technology focus.
  • Students will be encouraged to develop "bilingual" skills: that is, to study computing and another discipline together.
  • The university has already raised $650 million on the way to its $1 billion goal.

The biggest donation comes from Blackstone Group CEO Stephen Schwarzman, whose name will go on the college. In an interview with the New York Times, Schwarzman said one inspiration for the donation was his view that the U.S. has been "lagging" behind China on AI investments.

Flashback: MIT’s last big change was in the 1950s, when it added its management school and school of humanities.

  • Coming out of World War II, the university decided it needed to produce more well-rounded students, MIT Provost Martin Schmidt told Axios.
  • He sees a shift of that scale taking place now.
  • "Today, what we're experiencing is basically every discipline on this campus being transformed by advanced computational capabilities," Schmidt said.

What to watch: When asked if AI ethics — a major focus of MIT’s announcement — might be made mandatory, Schmidt said it would "make a lot of sense" to make ethics a strong focus in the new college.

  • One goal is to get more faculty in various disciplines thinking about ethical issues, said Melissa Nobles, dean of MIT’s School of Humanities, Arts, and Social Sciences.
  • Faculty in other departments leery of the societal effects of technology and AI might use the connections with the computing college to critique the work of technologists, said Nobles.

Go deeper:

4. Worthy of your time

Data: Council of Europe (2006), NIPSSR (2017), Eurostat (2018), and Martin et al. (2018) via United Nations Population Fund; Note: Final values for Russia and Japan are from 2015; Chart: Harry Stevens/Axios
Data: Council of Europe (2006), NIPSSR (2017), Eurostat (2018), and Martin et al. (2018) via United Nations Population Fund; Note: Final values for Russia and Japan are from 2015; Chart: Harry Stevens/Axios

This house explains how to adapt to climate change (Patricia Mazzei — NYT)

New norm: Babies out of wedlock (Stef Kight — Axios)

Report: AI is crucial to all emerging tech (Steve Rosenbush — WSJ) (subscription)

A risky, cutting-edge Walmart (Alistair Gray, Pan Kwan Yuk — FT) (subscription)

Insurers up premiums in California fire traps (Debra Kahn — Scientific American)

5. 1 fun thing: China infiltrates your local bar

Liquor bottles hanging from strings outside a store

Bottles of baijiu on display in Beijing. Photo: Liu Jin/AFP/Getty Images

As booze drinking declines in the West, companies are banking on cannabis-infused drinks and snacks to win the day. But the East is making a different bet.

What's happening: In China, the most popular alcohol is baijiu (白酒), a strong spirit made from fermented sorghum that raked in over $100 billion in sales last year. Compare that to whiskey and vodka, which earned about $40 billion each.

  • Baijiu ranges from cheap to very fine. In fact, Alibaba's finance arm, Ant Financial, is working on a blockchain to trace and verify the authenticity of Maotai, a particularly pricey brand of the stuff.
  • Now, Luzhou Laojiao Co., a Chinese baijiu distiller, wants to try its luck in the U.S. and Europe, reports WSJ. In collaboration with Western investors, Luzhou — a state-owned company worth $9 billion — is backing Ming River baijiu, a brand that sells the Chinese liquor to a global market at $40 a bottle.
  • It's already starting to pop up in New York City and London bars.