AI Now? More Like AI No How, According to That New York University Group’s Highly Political 2019 Report

When I got a message containing an annual summary of what happened this year in artificial intelligence from an organization I thought to be only an academic think tank, I was expecting to relay the good and the bad, along with many things of which neither of us were aware.  I printed its 59 pages before the endnotes and only later looked it over.


That proved to be just about a waste of paper, which, unfortunately, would have concerned the 2019 Report’s authors more than any lives AI has saved or improved.  When I started reading I found not the summary I had expected, but another one.  The AI Now Institute is not quite what I had thought – true, it is an “interdisciplinary research institute,” but it is “dedicated to understanding the social implications of AI technologies.”  That does not mean, though, that the group wanted to know about, say, the effect of Alexa devices on dinnertime conversations.  Nowhere in the text did I see anything positive about AI’s effects, and when it wandered off their official purpose, which was usually into the subject of regulation, that did not change.  


The Executive Summary had five bullet points.  To start, “the spread of algorithmic management technology in the workplace is increasing the power asymmetry between workers and employers.  AI threatens not only to disproportionately displace lower-wage earners, but also to reduce wages, job security, and other protections for those who need it most.”  True, but the same goes for globalization, efficiency, and other forms of automation, which as I have written are trends unlikely to be reversed or even halted.  


The second point was “community groups, workers, journalists, and researchers – not corporate AI ethics statements and policies – have been primarily responsible for pressuring tech companies and governments to set guardrails on the use of AI.”  In the short term, yes, but only because larger organizations are slower to respond.  Third, “efforts to regulate AI systems are underway, but they are being outpaced by government adoption of AI systems to surveil and control.”  Yes, Washington has plenty of use for the technology, and here, as elsewhere, not only AI itself but policies around it are evolving as quickly as they can.  


The fourth, “AI systems are continuing to amplify race and gender disparities via techniques like affect recognition, which has no sound scientific basis,” gets to the heart of what AI Now seems to be about, which is complaining about outcome differences between blacks and whites and men and women.  All involved with AI would agree that affect recognition, described here as “a subset of facial recognition that claims to “read” our inner emotions by interpreting the micro-expressions on our face,” is in its infancy, and its being “deployed at scale” in employment interviews, if that is reasonable to say, would be more objectionable if interviews and resumes were not notoriously weak and, yes, similarly flawed, without it.  In another year, leading-edge research on affect recognition will almost certainly point to methods different from those now in use.  


Last, “growing investment in and development of AI has profound implications in areas ranging from climate change to the rights of healthcare patients to the future of geopolitics and inequities being reinforced in regions of the global South.”  It is a stretch to call computer applications especially bad for the environment, and saying that “training just one AI model,” which can take months, produced as much carbon dioxide as “125 round-trip flights from New York to Beijing” did not refute that.  Barring interconnections between health care information systems can cost not only convenience and money but lives, and “the global South,” whatever that is, will not always be left alone.  


The 2019 Report also contains 12 recommendations, even sharper and more limiting than the main ideas above.  They include: banning affect recognition “in important decisions that affect people’s lives and access to opportunities” (in other words, whenever it is used in production); the need to “halt all use of facial recognition in sensitive social and political contexts until the risks are fully studied and adequate regulations are in place” (determination of when these two conditions are met, presumably, left to the likes of the AI Now Institute); requiring “the AI industry” to “make significant structural changes to address systemic racism, misogyny, and lack of diversity,” apparently with meanings of those terms mandated, all protected groups represented at 100% of their population shares, and all by-group pay levels exactly equal; several recommendations already in progress; references to “historic injustices,” the undefinable “social harms,” and “diverse cultural approaches to health” (most likely only along black-white, male-female, and maybe gay-straight lines).  


For one of many examples of a truer picture of AI, look at David Brooks’s June 24^th New York Times “How Artificial Intelligence Can Save Your Life.”  Though a short commercial journalistic piece, it told us that “researchers… were able this year to hook people up to brain monitors and generate natural-sounding synthetic speech out of mere brain activity” (and I doubt that would work only for white straight men) and that analyzing words used by people texting a suicide help line would greatly augment assessing their needs for emergency interventions (tendencies likewise valid for everyone), and concluded by conceding that “you can be freaked out by the privacy-invading power of A.I. to know you, but only A.I. can gather the data necessary to do this,” and “if it’s a matter of life and death, I suspect we’re going to” consent to such information use.  That is an even-handed conclusion, which, public-university knowledge-seeking auspices or not, the AI Now Institute, with its other “most recent publications” titled the likes of “Discriminating Systems:  Gender, Race, and Power on AI,” does not seem to offer.  That should tell us something.

Understanding Guaranteed Income and The Quality of Our Jobs

I’m surprised to see how long it has been since I wrote on universal basic income (UBI).  I made the main points about it long ago:  it would be expensive but much of its cost would be covered by removing welfare-related programs; it is not the same as unemployment benefits; while we don’t need it now, we must address it soon.  After reading the New York Times July 15^th, 2018 book review front-page article, “Barely Afloat in America” with two reviews of three books by Emily Cooke and former labor secretary and author Robert B. Reich, what do I think has changed?


The books themselves, Annie Lowrey’s Give People Money, Alissa Quart’s Squeezed, and current Democratic presidential candidate Andrew Yang’s War on Normal People, seem accurate and even visionary.  The first and last looked at the merits of guaranteed income, and Quart’s considered deterioration of the American middle class.  


Reich saw, and I see, a jobs situation worse than current unemployment figures.  On why, though, we part company.  I think that what we have is a temporary reprieve from what will, with the next recession, clearly show itself as a permanent problem of simply not enough work opportunities.  Reich said that there was no such thing, but rather “a good jobs crisis” (italics his).  He correctly pointed out that “most economic gains are going to a few,” said the solution was a UBI, which “may seem like a pipe dream” but “seems inevitable,” and cited both books as naming $1,000 per month – the same figure, though slightly smaller due to inflation, that I named in 2012’s Work’s New Age.  Per Reich, after me, that level “doesn’t deliver a comfortable life,” meaning most or at least many would still want paid employment, that about two-thirds of its $3.9 trillion annual cost would be covered by discontinuing food stamps and the like, and gave high billing to a financial transaction tax toward that shortfall.  Reich also, accurately and rarely, pointed out that once such payments are restricted to people with certain employment, income, or net worth statuses, or even taxed on a sliding scale, they are “no longer universal or even basic.”  At the end, though, he couldn’t resist mentioning “vastly unequal economic and political power” as a problem in itself, which UBI would not resolve.  


A more recent piece from Yang, “Yes, Robots Are Stealing Your Job,” in this year’s November 14^th New York Times, made the case that, despite recent downplaying, mechanization is still with us.  It is more or less a candidate’s position paper, but I can’t denigrate it for that, since where are his opponents’?  He also touted a $1,000 UBI as one of several measures to improve that situation, and ended with a call to “talk about the real causes and solutions of these problems instead of blaming the current occupant of the Oval Office.”  


I share a problem with Yang.  We are both stuck in the future.  That can be good or bad, positive when, as here, it draws advance attention to something worthy of it, but negative when it confuses what we need to do now with what does not require immediate action.  It can also draw a sort of Cassandra effect, in which when something doesn’t materialize people think it never will, and it taints subsequent predictions’ credibility.  Yet we need reasonably bipartisan forecasters.  


The other part of the Times Book Review piece, a look at Squeezed, was juxtaposed with Cooke’s rather dramatically expressed personal experiences, which led us to think there are day-care centers where the employees let their toddler charges run free while they look for TaskRabbit income supplementing opportunities, that foreign women “must” move away from their own children, and that she herself, despite working as “the editorial director of the New Republic,” does not expect to be able to afford more than one child.  The key thing to think about here, though, is another I put forth eight years ago, that we cannot expect to have the same things our parents did – some have become unaffordable, while others are much cheaper and more accessible now.  Another important fact is that, yes, children are more expensive even in inflation-adjusted terms than they were before.  


Overall, we have the freedom to choose what is and is not worthwhile for us.  And stopping education at a low level is a luxury we can afford less than previous generations could.  UBI or not, we must adapt – that, as with the previous piece, is a message we must take home.

November Jobs Data: AJSN Edges Below 15 Million Latent Demand, While BLS Numbers Show Small but Real Improvements

I wasn’t sure about this morning’s Bureau of Labor Statistics Employment Situation Summary.  This edition, from November, could have been much the same as October’s, and therefore overall not much better or worse than September’s, with a published estimate of 180,000 net new nonfarm positions ending up overly optimistic.  But none of this happened.


The actual number of new jobs as above came in at 266,000, with small improvements to the previous two months putting us at a three-month 205,000 average.  The unadjusted unemployment rate was down 0.1% to 3.5%, with the unadjusted one staying at 3.3%.  The total number of officially jobless Americans lost 100,000 to 5.8 million, as did the count of those out for 27 weeks or longer, now at 1.2 million, and the number working part-time for economic reasons, which fell to 4.3 million.  The two measures of how common it is for our country-people to be working, the labor force participation rate and the employment-population ratio, split between unfavorable and neutral, with the former down 0.1% to 63.2% and the latter holding at 61.0%.  Average nonfarm payroll wages had the catch-up month they needed, combining a 7-cent gain with a 4-cent positive adjustment for October to reach $28.29.  


The American Job Shortage Number, the measure showing how many additional positions could be quickly filled if all knew it would be easy to get one, fell below 15 million for the first time during its tracking, but with an improvement of only a scattered 24,000, as follows:

Compared with a year ago, the AJSN is down 530,000, the improvement headed by 188,000 lower latent demand from those unemployed, followed by 115,000 fewer apiece from people discouraged and not looking for the past year, and the rest split between five other categories.  Only 32.6% of the AJSN now comes from those officially jobless.  


How good then, overall, was November?  I like the broad-basedness of its advances, when, with unemployment rates not leaving much room for improvement, any gains would be good.  Therefore, the turtle, after resting in October, took another step forward this time. 

America Isn’t Ending, But It Is Evolving

An article in the December Atlantic considered whether “a tectonic demographic shift” will manage to ruin the United States.  Yoni Appelbaum’s “How America Ends,” which printed out to 18 pages, started with the observation that “democracy depends on the consent of the losers,” and named changes we are experiencing, including that “red and blue areas have become more deeply hued” or more polarized, that political differences are now concerning parents about their children’s spouses more than religious or racial ones, and that more on each side are dehumanizing those on the other.  If you do as Appelbaum apparently did and count all Hispanics as nonwhite, you find that white Christians are no longer most voters.  He also offered the thought-provoking view that “the Republican Party has treated Trump’s tenure more as an interregnum than a revival,” and that the GOP is headed for trouble when he leaves office, which, as “conservatives… are losing faith in their ability to win elections in the future” means that our democracy could weaken.  The Republican National Committee matched that view, issuing a report between 2012 and 2016 calling for the party to cater more to Hispanics, Asians, blacks, Indians, women, and young people, who combined for nearly 75% of 2012 voters.  


Per Appelbaum, though, these are hardly the worst times our republic has seen.  The South, both before and after the Civil War, chose “countermajoritarian politics” instead of seeking support for measures from more than half of the population.  Political-related violence, while more common now than a couple of decades ago, occurred much more often then, with House of Representatives members, in their own chamber, threatening each other with guns.  Neither is the composition of the Supreme Court more uneven than ever, as it “by the 1850s had a five-justice majority from slaveholding states” – even the most avid Trump supporters do not expect 7 of the 9 justices to share their opinions.  And immigration was more restricted in the 1920s than Republicans have recently seriously proposed.  


In the meantime, Democrats have become the party of the establishment, and, along with it, the side with the wealthiest participants.  This is hardly the first time Democrats and Republicans have switched their constituencies, and points more to change than termination.  


So if Republicans are indeed in long-term trouble, how can they escape?  Here are eight ways, all of which I, as a registered one, would like to see.  First, the party needs to overtly appeal to all ages, races, religions, and ethnic groups.  Second, it needs to emphasize common sense in social issues, in such ways as backing marijuana legalization and requiring equal rights between the sexes but needing more than a momentary self-identification to use bathrooms not matching chromosomes.  Third, as our president has totally failed to do, the party platform should emphasize prudent spending while not shorting safety-net programs.  Fourth, it would profit from finding and opposing Democrat-backed views, such as heavy laws and spending on climate change, with which many middle Americans disagree.  


Fifth, Republicans should support American interest while realizing that the country is indeed changing.  Sixth, they should strive to support as many jobs as possible without mandated- benefit or high-minimum-wage encumbrances.  Seventh, it must go out of the way to show that blacks, gays, and Hispanics, to name three, are more than welcome, not only as voters but as organizers and party decision-makers.  


Eighth, when in doubt, the GOP should choose a libertarian bent.  In the article above, Appelbaum quoted political scientist Adam Przeworski saying that our party system “must give all the relevant political forces a chance to win from time to time in the competition of interests and values.”  Libertarians, of which many others are at least partial supporters, have not had theirs yet.  Freedom is as American as the Super Bowl – why not try more of it?  That, beyond almost anything else, is what our evolving, not dying, country should try. 

Driverless Cars – In the Metal!

Eight days ago I sought out and visited Chandler, Arizona.  With 65 square miles and over a quarter million people, that Phoenix satellite is one of the newer huge suburbs, such as Henderson, Nevada and Plano, Texas, which grew by expanding into unincorporated land while adding few features of standalone cities.  It seemed like a fine place to live, but why go there?

The answer, which regular readers of this blog may have suspected even before reading today’s headline, is that Chandler has become perhaps the most important place for the real-life road-testing of autonomous vehicles.  I had known they were common there, but despite writing tens of thousands of words on that subject had never actually seen one.  But now I have.  So you yourself don’t need to try to photograph moving cars from other ones, I chased them around and shot them on my trusty Canon, and returned with the following 13 things to say.

First, in about an hour’s searching I saw 8 to 10 autonomous Waymo vehicles, all moving, and which could have included duplicates.  Second, two seemed to be on courses, making the same sequence of turns.  Three, each had one or two on-board safety drivers, and one I got next to also had a, rather bored-looking, back seat passenger.

Four, the cars acted normally in traffic and actually exceeded the conservative posted speed limits a bit; I almost lost one when it went over 50 for almost a mile in a 45 zone.  Five, most of the vehicle tops had cylinder-shaped rooftop Lidar devices, though some had extra thick flat tops.  Six, all were white sort of small, longish SUVs with the faint turquoise Waymo name.  Seven, I saw not a single out-of-the-ordinary reaction, including jeers or vandalism attempts, from other drivers or pedestrians – all acted as if the vehicles were nothing special at all.

Eighth, I talked with (actually at) two safety drivers, encouraging them and acknowledging that while I was being a pest I was on their side – they, men in their 20s or around 30, seemed cheerful, sober sorts.  Ninth, Chandler did look new and prosperous, with streets all in excellent condition.  Tenth, at a sunny or slightly cloudy 80 that afternoon the weather was perfect.

Eleventh, Chandler’s larger roads followed a clear grid pattern, but others, often running through subdivisions, wiggled around.  I saw two expressways, one on the western border and another going through the middle of town, with orderly exit patterns.  Twelfth, except for one where the safety driver was using the wheel to back out of a parking space, I could not tell when the cars were in autonomous mode.

Thirteenth, we can contrast the unobtrusiveness of these vehicles with the fuss the horseless carriages made in their early appearances over 100 years ago.  The equivalent of the Clancy Brothers singing that “you could hear the din all through Glenfin of Johnston’s motor car” will not happen this time.  After a while, we will not notice them – and that is the way that should be.

Here are the photos:

Autonomous Vehicles: The Hits Just Keep On Coming

Three months ago I posted on why self-driving car progress was stalling out, and how we could get it moving again.  Since then, the situation has changed – it’s become worse.

Soon after that date, Clyde Haberman’s July 14^th New York Times “Driverless Cars Are Taking Longer Than Expected.  Here’s Why.,” offered a recap of a few of the technology’s problems:  insufficient mapping, which must be done “down to a few centimeters”; remaining issues with bad weather; the perception problem worsened by Boeing’s remarkably recalcitrant 737 Max troubles; and remaining fear from the sole pedestrian death.  There are also real and widespread legal issues, as described in David Shepardson’s August 29^th Reuters “Waymo urges U.S. to ‘promptly’ remove barriers to self-driving cars,” including the need to “meet nearly 75 auto safety standards for self-driving cars, many of them written under the assumption that a licensed driver is in command of the vehicle using traditional controls,” which the National Highway Traffic Safety Administration, which may not “complete a comprehensive rewrite of various safety standards” before 2025, will address in March.  Shepardson’s piece implicitly makes a case for development in limited areas, as is happening to a great extent now, to be, if successful, propagated after that.

More discouraging though was Brett Berk’s November 3^rd Car and Driver “The Enemies of the Autonomous Vehicle.”  I’m glad to see that publication, which may someday need to drop the two last words from its name, weighing in here.  Berk started with the news that Chandler, Arizona, long a center for driverless cars, has seen “nearly two dozen” attacks, with people “pelting them with rocks, trying to run them off the road, challenging them to games of chicken, and slashing their tires.”  Some of that we can expect, by rather young perpetrators acting out of curiosity, and Berk saw no evidence of any organized revolt. 

The author, though, found more than that to cause concern.  An “earlier this year” AAA survey still showed that about 75% feared riding in autonomous vehicles.  A man, citing the pedestrian death, threatened one with a .22 pistol.  As likely, given the lack of widespread rollouts, “no national or industry standards currently exist for the regulation of these vehicles,” beyond banning them entirely as before.  And while the Chandler troubles were scattered, organizations named Save Driving and The Human Driving Association, the latter with almost 10,000 people, have popped up and are voicing a range of objections, including that self-driving cars would “enhance national divisions” between rural and urban and would provide new masses of data to the huge companies providing them.  There is also union opposition, some remarkably enlightened, with Teamsters president James Hoffa Jr. calling out autonomous vehicle developers for wanting above all to spend less money, and asking that those savings be used to help drivers whose jobs it has cost.  As well, the Cadillac Super Cruise partial automation now for sale does not consistently work, and Berk echoed Haberman’s observation that rain and snow still cause plenty of problems.

Then, as maybe the worst problem of all, we have hacking.  I still maintain that it can be beat, but its variety, sophistication, and technical intensity mean that will require a great deal of time and funds.  Per Berk, a benevolent Chinese research group found that “small stickers placed in a roadway were enough to convince the (self-driving) car to change lanes right into oncoming traffic.”  Overall, “cars could be taken over remotely and manipulated, militarized, or held hostage for ransom.”  As such vehicles have over 300 million lines of code, they have more vulnerable places than commercial planes with their 15 million.

For those of us in favor of automated vehicles, all of this is sad.  And unless research efforts regain their passion, and plenty of other stakeholders help out and clear the way, their state in 2025 won’t look much different from now, with, among other losses, another 150,000 dead Americans.  That would be even more sorrowful.

Jobs Report Barely Moved for October, A Good Thing Given Forecasts – AJSN Latent Demand Down 100,000 to 15.0 Million

This morning’s Bureau of Labor Statistics Employment Situation Summary was supposed to be a stinker, or at least only somewhat unfavorable.  Did that happen?

I saw four forecasts of the recent-marquee number of net new nonfarm positions, all lower than last month’s.  The result beat them all.  Instead of from 75,000 to 125,000 we had 128,000, not impressive but about enough to cover population growth.  Elsewhere, significant results were few.  The seasonally adjusted number of unemployed gained 100,000, offset by the labor force participation rate up 0.1%.  Official adjusted unemployment rose 0.1% to 3.6%.  Wages gained 9 cents per hour over September’s first report, which was later revised slightly upwards leaving us with a 6-cent October rise, just about inflation, to $28.18.  The others on which I have been reporting, the unadjusted jobless rate (3.3%), the count of people in that category for 27 weeks or longer (1.3 million), the employment-population ratio (61.0%), and those working part-time for economic reasons, or keeping such positions while thus far unsuccessfully seeking a full-time one (4.4 million), all broke even.

The American Job Shortage Number or AJSN, the metric showing how many more positions could be quickly filled if getting one were as easy as getting a pizza, reached another multiyear low, but it was close – it improved only almost 100,000, as follows:

The largest change was from a 138,000 improvement in the number of people claiming interest in working but not looking for a year or more.  The 45,000 increase in those unemployed offset that by 40,500, and changes beyond that were all 18,000 or lower.  The share of the AJSN from people officially jobless gained a little bit to 33.1%, still under one-third. 

Compared with a year before, the AJSN is still getting better.  October 2018’s result was 700,000 higher at 15.7 million, almost all due to higher unemployment, more not searching for a year or longer, and a rise in those calling themselves discouraged.

How did we do overall?  The good news is that this was not the down month many expected.  The bad news is that we didn’t go anywhere positive either.  There is still no sign of recession here, but although I saw the turtle move his knee, he stayed right where he was.