Ways to think about machine learning

 

I think one could propose a whole list of unhelpful ways of talking about current developments in machine learning. For example:

  • Data is the new oil
  • Google and China (or Facebook, or Amazon, or BAT) have all the data
  • AI will take all the jobs
  • And, of course, saying AI itself.

More useful things to talk about, perhaps, might be:

  • Automation
  • Enabling technology layers
  • Relational databases.

.. Before relational databases appeared in the late 1970s, if you wanted your database to show you, say, ‘all customers who bought this product and live in this city’, that would generally need a custom engineering project. Databases were not built with structure such that any arbitrary cross-referenced query was an easy, routine thing to do. If you wanted to ask a question, someone would have to build it. Databases were record-keeping systems; relational databases turned them into business intelligence systems.

This changed what databases could be used for in important ways, and so created new use cases and new billion dollar companies. Relational databases gave us Oracle, but they also gave us SAP, and SAP and its peers gave us global just-in-time supply chains – they gave us Apple and Starbucks.

.. with each wave of automation, we imagine we’re creating something anthropomorphic or something with general intelligence. In the 1920s and 30s we imagined steel men walking around factories holding hammers, and in the 1950s we imagined humanoid robots walking around the kitchen doing the housework. We didn’t get robot servants – we got washing machines.

.. Washing machines are robots, but they’re not ‘intelligent’. They don’t know what water or clothes are. Moreover, they’re not general purpose even in the narrow domain of washing – you can’t put dishes in a washing machine, nor clothes in a dishwasher
.. machine learning lets us solve classes of problem that computers could not usefully address before, but each of those problems will require a different implementation, and different data, a different route to market, and often a different company.
.. Machine learning is not going to create HAL 9000 (at least, very few people in the field think that it will do so any time soon), but it’s also not useful to call it ‘just statistics’.
.. this might be rather like talking about SQL in 1980 – how do you get from explaining table joins to thinking about Salesforce.com? It’s all very well to say ‘this lets you ask these new kinds of questions‘, but it isn’t always very obvious what questions.
  1. .. Machine learning may well deliver better results for questions you’re already asking about data you already
  2. .. Machine learning lets you ask new questions of the data you already have. For example, a lawyer doing discovery might search for ‘angry’ emails, or ‘anxious’ or anomalous threads or clusters of documents, as well as doing keyword searches,
  3. .. machine learning opens up new data types to analysis – computers could not really read audio, images or video before and now, increasingly, that will be possible.

.. Within this, I find imaging much the most exciting. Computers have been able to process text and numbers for as long as we’ve had computers, but images (and video) have been mostly opaque.

.. Now they’ll be able to ‘see’ in the same sense as they can ‘read’. This means that image sensors (and microphones) become a whole new input mechanism – less a ‘camera’ than a new, powerful and flexible sensor that generates a stream of (potentially) machine-readable data.  All sorts of things will turn out to be computer vision problems that don’t look like computer vision problems today.

.. I met a company recently that supplies seats to the car industry, which has put a neural network on a cheap DSP chip with a cheap smartphone image sensor, to detect whether there’s a wrinkle in the fabric (we should expect all sorts of similar uses for machine learning in very small, cheap widgets, doing just one thing, as described here). It’s not useful to describe this as ‘artificial intelligence’: it’s automation of a task that could not previously be automated. A person had to look.

.. one of my colleagues suggested that machine learning will be able to do anything you could train a dog to do

..  Ng has suggested that ML will be able to do anything you could do in less than one second.

..  I prefer the metaphor that this gives you infinite interns, or, perhaps, infinite ten year olds. 

.. Five years ago, if you gave a computer a pile of photos, it couldn’t do much more than sort them by size. A ten year old could sort them into men and women, a fifteen year old into cool and uncool and an intern could say ‘this one’s really interesting’. Today, with ML, the computer will match the ten year old and perhaps the fifteen year old. It might never get to the intern. But what would you do if you had a million fifteen year olds to look at your data? What calls would you listen to, what images would you look at, and what file transfers or credit card payments would you inspect?

.. machine learning doesn’t have to match experts or decades of experience or judgement. We’re not automating experts. Rather, we’re asking ‘listen to all the phone calls and find the angry ones’. ‘Read all the emails and find the anxious ones’. ‘Look at a hundred thousand photos and find the cool (or at least weird) people’.

.. this is what automation always does;

  • Excel didn’t give us artificial accountants,
  • Photoshop and Indesign didn’t give us artificial graphic designers and indeed
  • steam engines didn’t give us artificial horses. ..

Rather, we automated one discrete task, at massive scale.

.. Where this metaphor breaks down (as all metaphors do) is in the sense that in some fields, machine learning can not just find things we can already recognize, but find things that humans can’t recognize, or find levels of pattern, inference or implication that no ten year old (or 50 year old) would recognize.

.. This is best seen Deepmind’s AlphaGo. AlphaGo doesn’t play Go the way the chess computers played chess – by analysing every possible tree of moves in sequence. Rather, it was given the rules and a board and left to try to work out strategies by itself, playing more games against itself than a human could do in many lifetimes. That is, this not so much a thousand interns as one intern that’s very very fast, and you give your intern 10 million images and they come back and say ‘it’s a funny thing, but when I looked at the third million images, this pattern really started coming out’.

.. what fields are narrow enough that we can tell an ML system the rules (or give it a score), but deep enough that looking at all of the data, as no human could ever do, might bring out new results?

Trump is so obsessed with winning that he might make America lose

In his zero-sum universe, you’re either victorious or you’re defeated.

 “I win against China. You can win against China if you’re smart,” he said at a campaign event in July 2015.
.. “Vast numbers of manufacturing jobs in Pennsylvania have moved to Mexico and other countries. That will end when I win!”
.. “China, Japan, Mexico, Brazil, these countries are all taking our jobs, like we’re a bunch of babies. That will stop,”
.. In Trump’s view of the world, there is a finite amount of everything — money, security, jobs, victories — and nothing can be shared.
.. It’s a universe where the strong do what they can and the weak suffer what they must, as Thucydides said.
.. The problem is that the triumphs that Trump craves — strength, safety, prosperity — cannot be achieved alone.
.. They require friends and allies, and they require the president to see those people as partners, not competitors.
.. other governments don’t like to be punching bags, the only role he appears to envision for them.
.. In real estate, relationships often take the form of one-off transactions: You can cheat people you’ll never do business with again.
.. Winners have trade surpluses, and losers have trade deficits.
.. The United States is the biggest economy with the biggest military, and therefore the United States has leverage to get the best deals. If we don’t emerge from negotiation with a clear advantage, that’s because our negotiator was a soft-headed, do-gooder globalist who didn’t put America first.
.. Washington has the most leverage when it deals with countries one on one, which is why, he says, “we need bilateral trade deals,” not “another international agreement that ties us up and binds us down.”
To abide by the same rules as less-powerful countries would be to sublimate American interests to those of lesser nations.
..  Trump seeks to begin negotiations with a threat that forces the other side to defend its smaller piece. He pledges to tear up NAFTA, rip up the Iran nuclear deal and revisit America’s relationship with NATO — unless he gets concessions.
.. he gains advantage not by telling the truth but by saying things he believes will boost his bargaining power and sell his vision: China has been allowed to “rape our country.”
.. He’s just an alliance-hating unilateralist.
..  he sees three kinds of immigrants:
  1. smart guys from smart countries, like Norway,
  2. undeserving charity cases from “shithole” countries and
  3. terrorists/gang members who threaten ordinary Americans.

.. The zero-sum cosmology touches everything. Obamacare supposedly sticks us with the bill for people who should pay for their own insurance — or a find a job that provides it.

..  he doesn’t exercise, because “the human body was like a battery, with a finite amount of energy, which exercise only depleted.”

..  China is more a strategic competitor than a real partner linked by shared values.

.. “after more than four decades of serving as the nation’s economic majority, the American middle class is now matched in number by those in the economic tiers above and below it.” That’s a real problem, and Trump is right to point it out.

.. He could have demanded that NATO members pay more without signaling that he might abandon the mutual-defense agreement that undergirds a treaty to contain Russia.

.. Relations among nations are not like real estate deals. The president has to negotiate with the same people again the next month, and they’ll remember how they’ve been treated.

Israel’s Benjamin Netanyahu never forgave President Barack Obama for openly criticizing his approach to settlement-building;

imagine how every other leader feels about being constantly humiliated by Trump.

.. Other countries form judgments about whether American promises are credible and whether they can trust the president. Trump says he’s willing to talk with North Korea about its nuclear program, but surely Kim Jong Un is watching as Trump threatens to shred the Iran nuclear agreement.

..  The Belt and Road Initiative, China’s plan to blaze new commercial trails and cement new political ties via infrastructure investment in dozens of countries, is seven times larger than the Marshall Plan when adjusted for inflation.

.. More than 120 nations already trade more with China than with the United States.

.. China is investing in smaller European Union members like Hungary and Greece to alter official E.U. attitudes toward Beijing. That’s why the Trans-Pacific Partnership, which Trump quashed, was more than just a trade deal. By joining, Trump could have expanded U.S. ties with many of China’s neighbors, governments that fear overreliance on China’s goodwill for future growth.

.. Trump’s win-or-lose philosophy is most confused when it comes to immigration. Foreigners who want to become Americans are not charity cases. They participate in the labor force at higher rates (73.4 percent, according to the Bureau of Labor Statistics) than native-born Americans.

.. Trump’s tendency to hire foreign guest workers over Americans at his own properties suggests that he understands something about how hard they work.

.. The undocumented contribute $13 billion to the nation’s retirement fund each year and get just $1 billion in return.

.. “More than three out of every four patents at the top 10 patent-producing US universities (76%) had at least one foreign-born inventor,”

..  tourism has fallen 4 percent, with a resulting loss of 40,000 jobs. Foreign applications to U.S. universities are down, too.

.. he doesn’t seem to know that some of our country’s greatest success stories began in failure.

  1. Thomas Edison famously erred 1,000 times on the way to inventing the light bulb — it “was an invention with 1,000 steps,” he said.
  2. Henry Ford went broke repeatedly before he succeeded.
  3. Steve Jobs, a college dropout, was fired from the company he founded. Even
  4. Trump’s own businesses have gone bankrupt.

.. If he wants to track terrorists before they try to enter the United States, he needs support from foreign intelligence services.

.. Today, the United States doesn’t have that kind of leverage, and Trump’s aggressive criticism of other countries, including allies, poisons public attitudes toward the United States and makes it harder for foreign leaders to cooperate with Washington publicly.

.. Trump and his leadership at some of the lowest levels since Pew began tracking the U.S. image abroad in 2002. Almost three-quarters of those surveyed said they have little to no confidence in Trump.

.. if Trump wants to make the best deals, he’ll need to learn a few words:

  1. respect,
  2. cooperation and
  3. compromise.

These ideas won’t fire up a campaign rally. But they might help build an American strategy that works.

 

 

The Economics of Dirty Old Men

About washing machines: The legal basis of the new tariff is a finding by the United States International Trade Commission that the industry has been injured by rising imports. The definition of “injury” is a bit peculiar: The commission admitted that the domestic industry “did not suffer a significant idling of productive facilities,” and that “there has been no significant unemployment or underemployment.” Nonetheless, the commission argued that production and employment should have expanded more than it did given the economy’s growth between 2012 and 2016 (you know, the Obama-era boom Trump insisted was fake).

.. Everything we know about the Trump administration suggests that hurting renewables is actually a good thing from its point of view. As I said, this is an administration of dirty old men.

.. Over all, there are around five times as many people working, in one way or another, for the solar energy sector as there are coal miners.

.. Last fall, Rick Perry, the energy secretary, tried to impose a rule that would in effect have forced electricity grids to subsidize coal and nuclear plants. The rule was shot down, but it showed what these guys want. From their point of view, destroying solar jobs is probably a good thing.

.. what’s good for the Koch brothers may not be good for America (or the world), but it’s good for G.O.P. campaign finance. Partly it’s about blue-collar voters, who still imagine that Trump can bring back coal jobs. (In 2017 the coal industry added 500, that’s right, 500 jobs. That’s 0.0003 percent of total U.S. employment.)

.. It’s also partly about cultural nostalgia: Trump and others recall the heyday of fossil fuels as a golden age

.. But I suspect that it’s also about a kind of machismo, a sense that real men don’t soak up solar energy; they burn stuff instead.

.. You shouldn’t even call it protectionism, since its direct effect will be to destroy far more jobs than it creates. Plus it’s bad for the environment. So much winning!

Behind Iran’s protests, anger over lost life savings and tightfisted budgets

The unnamed woman is one of countless Iranians who say their savings have been wiped out by the collapse of fraudulent businesses and unlicensed credit institutions in recent years. Economists are now pointing to the abrupt closure of these poorly regulated institutions as laying the foundation for the unrest that struck Iran starting in late December.

.. “Banks are shutting down without any kind of notice, and it’s creating a huge political and economic backlash at a local level,” said Suzanne Maloney, senior fellow on Middle East policy at the Brookings Institution.

.. it seems to have tapped into a deep sense of alienation and frustration, that people aren’t just demonstrating for better working conditions or pay, but insisting on wholesale rejection of the system itself.”

.. the average budget of Iranian households declined by 15 percent from 2007 — when the U.N. Security Council adopted some of its toughest sanctions on Iran — to 2016.

.. Iranian President Hassan Rouhani, a relative moderate who was reelected to a second term in May, has carried out a program of fiscal austerity. It has brought down inflation but hurt job growth

.. Rouhani has also imposed what Salehi-Isfahani called “regressive policies,” such as raising energy prices while shrinking cash transfers that the poor use to pay for essential items.

.. Other new policies have favored businesses and the middle class, whose members predominantly reside in the capital, Tehran

.. Iran has seen a “divergence in living standards (measured by per capita expenditures) between Tehran on one side and the rest of the country on the other

.. The budget envisioned steep cuts for cash subsidies to the poor, while increasing fees for things like vehicle registration and traveling abroad.

.. Rouhani’s budget was also notable because it was the first time the government made public the funds allocated to Iran’s wealthy religious foundations — as well as its powerful military and paramilitary forces.

.. The disclosure of an $8 billion budget for the Revolutionary Guard Corps, Iran’s most influential security body, prompted sharp criticism from protesters who objected to government spending on Iranian involvement in regional wars, including in Iraq and Syria.

.. Religious foundations, many of which are tax exempt, also got a boost in the new budget, including, for example, a 20 percent increase for representatives of the supreme leader, Ayatollah Ali Khamenei, posted at Iran’s universities.

.. Rouhani sold the nuclear deal to Iranians as crucial for reviving the ailing economy. Iranians have been disappointed that growth has not been faster, including 74 percent who said in July that there had been no economic improvement as a result of the deal