Artificial Intelligence, Present Tense: Modest Revenues and Serious Problems

There is a massive amount of AI material online.  Almost all of it, though, deals with capabilities in development, business deals, stock performance, organizational changes, and forecasts.

How is it doing now?  I don’t know, and you don’t either.  It moves too quickly, and most of the information is proprietary and unassembled.  But if we can go back four months, to “10 Companies With Largest AI Revenues:  Success Cases from OpenAI, Anthropic & More” (Tim Keary, Technopedia, January 17^th), we get some insight.  In the first paragraph it relates how “the global AI market was valued at $196.63 billion in 2023,” close to Statistica’s October 2023 $207.9 billion determination.  However, per an unsourced answer to a Google search, “the “market size” is made up of the total number of potential buyers of a product or service within a given market, and the total revenue that these sales may generate.”  As Keary told us, though, that’s not the same as what’s actually being sold.  In his section titled “10 Leading Companies in Terms of AI Revenue,” he named OpenAI, Anthropic, Microsoft, Nvidia, Hugging Face, Stability AI, Perplexity AI, IBM, Google, and Salesforce, giving annual sales figures for each.  After taking one-quarter of those from the last two, which were not broken down into AI and other offerings, I got a total of $29.7 billion, meaning that something over 15% of the market is producing revenue.  By comparison, a recent Wikipedia list of largest companies by annual sales – not entire industries – showed the 50th-largest one with over $157 billion.

Accordingly, “A.I. Start-Ups Face a Rough Financial Reality Check” (Cade Metz, Karen Weise, and Tripp Mickle, The New York Times, April 29^th).  The main point here was that “the A.I. revolution… is going to come with a very big price tag.  And the tech companies that have bet their futures on it are scrambling to figure out how to close the gap between those expenses and the profits they hope to make somewhere down the line.”  Although “investors have poured $330 billion into about 26,000 A.I. and machine-learning start-ups over the past three years,” they may not maintain that pace.  Indeed, “Wall Street’s Patience for a Costly A.I. Arms Race Is Waning” (Andrew Ross Sorkin et al., New York Times DealBook Newsletter, April 25^th). 

Also, “A.I. is running out of power” (Andrew Ross Sorkin et al., New York Times DealBook Newsletter, April 18^th).  As a result, “tech executives are increasingly warning that electricity supplies need a boost,” with a notice that “there’s “not enough energy right now” to power new generative A. I. services,” and because of that the technology “could also spur a geographic shift for tech” to areas with more of it to spare.  With that, it is fair to ask “How Bad is A.I. for the Climate?” (the same source on May 6^th), as it “risks throwing companies’ climate pledges off track” since “the A.I. revolution will largely run on fossil fuels.”

Another copyright infringement lawsuit materialized last week, as “8 Daily Newspapers Sue OpenAI and Microsoft Over A.I.” (Katie Robertson, The New York Times, April 30^th).  These papers, including the New York Post and the Chicago Tribune, all owned by one company, claimed that “chatbots regularly surfaced the entire text of articles behind subscription paywalls for users and often did not prominently link back to the source,” and erroneously named the papers as sources of things they did not produce.

Although many predictions show massive increases in profits as well as capabilities, “CEOs make the message clear on AI’s big payoff:  Be patient” (Lloyd Lee, Business Insider, April 25^th).  To prevent AI-company stock prices from reverting to whence they came, when “many (companies) have yet to see any significant returns on their investment,” “CEOs hope to re-assure shareholders that this is to be expected.”  Venture capitalists have not always been known for patience, so even if the titanic success so often forecast for AI materializes, without every firm running out of power, data, chips, perceived accuracy, or legal acceptance, many may fail due to running out of money. 

We don’t know where or how far artificial intelligence is going, but it hasn’t done much yet.  From here, there are no guarantees – and plenty of doubts.

The Jobs Report Dips from Great to Good – AJSN Reflects Only Half of Its Seasonally Lower Latent Demand

This morning’s Bureau of Labor Statistics Employment Situation Summary may have seemed disappointing, as for once the number of net new nonfarm payroll positions fell short of public estimates.  But that was not the real reason.  The ones I saw for April, the month of the report, ranged from 223,000 to 250,000 – it came in at 175,000.  Otherwise, seasonally adjusted unemployment was up 0.1% to 3.9% and the unadjusted version was down, as is typical from March to April, 0.4% to 3.5%.  Unadjusted employment rose 234,000 to 161,590,000, with such unemployed down more than 700,000 to 5,894,000.  There were 95,080,000 people claiming no interest in work, up 266,000.  The counts of people jobless for 12 months or more and those working part-time for economic reasons, or keeping those arrangements while looking thus far unsuccessfully for full-time ones, both worsened, up from 1.2 million to 1.3 million and from 4.3 million to 4.5 million.  The two measures showing how common it is for Americans to be working or that plus those counted as jobless, the employment-population ratio and the labor force participation rate, were narrowly mixed, with the former down 0.1% to 60.2% and the latter holding at 62.7%.  Private nonfarm payroll hourly earnings increased 6 cents to $34.75, not keeping up with inflation.

The American Job Shortage Number or AJSN, the metric showing how many additional positions could be quickly filled if all knew that getting one would be little more than another daily errand, improved with a 300,000 drop from March as follows:

The number of unemployed shaved over 600,000, but increases in those discouraged and those not looking for a year or more offset more than half of that.  The share of the AJSN from those officially jobless fell to 33.1%, meaning that more than two-thirds of people taking these opportunities and not working now would have other statuses.  Compared with a year before, the AJSN gained about 650,000, more than that from a higher number of unemployed.

We have areas of concern here.  Although the number of new jobs keeps on exceeding our population increases, joblessness is at the top of its nine-month range, with several other statistics worse than in March.  If they degrade again, we may need to accept that we are not in the top employment category anymore.  In the meantime, though, the turtle took a small step forward.

Four More Facets of Working from Home

The usual news on this issue has settled down.  The pendulum which swings back and forth between office and remote labor has continued to move toward the former, with companies not in the news as much for telling their employees they can choose between turning up onsite or losing their jobs.  There have been some items on this topic, though, over the past ten weeks.

First was “The ‘work-from-home weekend’ may be on its way out” (Lakshmi Varanasi, Business Insider, February 16^th).  The main example was Deutsche Bank, which now “won’t let employees work from home on both Fridays and Mondays.”  Decades ago, at AT&T, those were the most common days for that, which when combined with generally unconscious bosses and worker measures taken to impede contact, meant too many people were in effect working only Tuesday through Thursday.  Ostensibly, “the move was made to even out worker attendance across the week,” but as it is easy to try to solve attendance and productivity problems with policy, it was probably more than that.

Second, “The ZIP Code Shift:  Why Many Americans No Longer Live Where They Work” (Emma Goldberg, The New York Times, March 4^th).  Per then-breaking research, “many Americans now live roughly twice as far from their offices as they did prepandemic,” not across the country, but choosing couple-hour commutes, which they do less than five days a week and rarely if they are among the still 12% of workers exclusively remote.  That is bad news for those with businesses in downtown areas hoping that with Covid’s fading they would recover to where they were in the late 2010s – but can be a good choice for people wanting to live in another city for housing or personal reasons. 

Does anyone wonder why there has been no great remote-worker movement to isolated rural areas and small towns?  It may be “The hidden price of leaving a big city” (Aki Ito, Business Insider, April 15^th).  While “it may improve your quality of life,” you may need “good luck if you lose your job.”  As well, some making that choice have ended up “pining for the things they left behind, from culinary excellence to cultural diversity.”  Yet, although the author said “every week it seems like I see a new story about some former San Franciscan or New Yorker regretting their decision to leave,” population has been dropping in those metropolitan areas along with Los Angeles and Seattle.  It is important if not always determinative to consider how suitable new locations would be for continuing careers.

Finally, one office disadvantage you may not have considered.  In “Google’s worker firings show that the office actually isn’t a place to be yourself” (Business Insider, April 20^th), author Tim Paradis told us that “Google fired more than two dozen workers after they took part in sit-ins at offices in California and New York.”  For about ten years there has been a trend toward encouraging employees to “bring your whole self to work,” probably strengthened by blurred home-office borders, recreation as part of the workday, and Generation Z wishes.  However, it is still dangerous to talk about politics, let alone participate in it at the office, and other subjects would also invite unwanted controversy.  The concern here seems quite like something I read almost 50 years ago, when jobseekers were cautioned that management will only indulge talking about “real or imagined personal problems” up to a point, that “you can get fired,” which meant “so much” to the idea of coworkers being equivalent to a family. 

The issue of whether to work from an office is still important.  Expect more here – especially if the pendulum shows signs of stopping or reversing.

Three Employee Problems and One Big Employer Concern – Different but Similar

With new technology, follow-on effects from the pandemic, strong employment, and a new generation – Z – becoming established in jobs, it is no surprise that concerns around workers and their managements are evolving.  Except for those centering around putting in less effort, having multiple jobs, and the pendulum of office work to home work and back, what issues have reached the press?

For one, “Employees are demanding workplace resets” (Brock Dumas, Fox Business, August 31^st).  The piece covered a report from Edelman, which once a year issues the Edelman Trust Barometer.  The statement revealed that, of 7,000 employee recipients, 72% “said it is more important than ever that employers rethink what work means to employees,” with “career advancement,” “personal empowerment,” and “societal impact” each named by 71% to 83% of respondents.  While the first one is old, the second has not been stated as often as felt and the third has recently risen dramatically.  Sixty-one percent said they were “more likely to work for an organization where the CEO speaks publicly about the controversial issues they care about.”  In all, though, “workers across the board have expressed far greater trust in their employers than any other institution the survey asks about, which are business, non-government organizations, the media and the government.”

Yet there are other problems.  Per “Most people have an unhealthy relationship with work, study finds: ‘Huge opportunity’” (Erica Lamberg, Fox Business, September 21^st).  The surveyor this time was HP Inc, with “the first HP Work Relationship Index, a comprehensive study that explores employees’ relations with work around the globe.”  Only 27% of “knowledge workers” claimed a “healthy relationship with work.”  Lamberg recommended employees talk about it with their bosses, with important principles to “walk before you run” or expect less at the start, “it’s all about presentation” or framing intentions as “here is how you can get the very best quality of work out of me,” and “know your worth” by documenting accomplishments. 

A long-standing but timely one returns in “Almost half of Americans see automation replacing their jobs” (Rich Miller, Benefit News, August 21^st).  American Staffing Association found that in a survey, a share which almost doubled since a similar 2017 effort, and that “some three-quarters of those polled expect increased use of automation and artificial intelligence to lead to higher joblessness.”  With 2024 thus far a down year for AI, it would be worthwhile to see what a similar study would tell us now.

On the employers’ side, one concern affecting all of us has evolved.  On December 11^th, we saw that “Corporate America Is Testing the Limits of Its Pricing Power” (Jason Karaian, Jeanna Smialek, and Joe Rennison, The New York Times, December 11^th).  Talking first about 2022 and then about late 2023, “margins eased somewhat last year, but have recently recovered to levels that would have set records before the pandemic,” and “average margins in nearly every sector in the S&P 500 are running near or above 10-year highs, according to Goldman Sachs.”  A chart titled “Quarterly net profit margin of S&P 500 companies,” covering 2010 to 2023, showed, except for an early-pandemic dip, a choppy rise from just over 8% to 12%.  Further article findings were “there’s a focus on margins over market share” and “that’s a shift from post-2009 practice,” “companies learned they can charge more than they thought… but price sensitivity may return,” and “the ability to raise prices – or keep them high – may not last.”

In the four months since, inflation concerns have become more vocal, so overall demand may yet return to being more elastic.  Yet some ways of raising profit are getting bad press, at least from Christopher Beam, whose “Welcome to Pricing Hell” came out in Atlantic in April.  The author’s concern was about “the ubiquitous rise of add-on fees and personalized pricing,” which “has turned buying stuff into a game you can’t win.”  He started discussing how Wendy’s’ plan for “dynamic pricing” faced strong negative reactions, as have new fees, subscription plans, unbundled features, and the use of personal data to determine product cost.  Variable rates, common before the 1800s invention of the price tag, may have started with airplane tickets, spread to “airline-adjacent industries like hotels and cruise lines,” and proliferated from there.  In the 2010s, more and more industries added fees often unrevealed until just before payment time.  At the origin point, extra charges have reached the point where “the airlines raked in $33 billion from baggage fees, and even more from other ancillary fees like seat selection, meals, and in-flight Wi-Fi,” such tariffs becoming “a major driver of airline profits.”  Personal data often reaches retailers through apps, which can access purchase history, financial information, and much more, which “can be fed into machine-learning algorithms to generate a portrait of you and your willingness to pay.” 

Most of my fellow economists think personal pricing is fine.  They do not, however, like belatedly disclosed fees.  As the last article revealed, “even the CATO Institute, the libertarian think tank that never saw a regulation it liked, acknowledges that consumers “shouldn’t be charged for products without their consent, and businesses should disclose mandatory fees before purchases are made.””  If that means sooner than the Place Order page, that’s good enough for me.  There will be much more personal pricing, though – it is up to us to determine how to facilitate or impede it, as we wish.  In work situations as well as with what we buy, there are countermeasures.  Even in these times, we still have choices.

Artificial Intelligence: Twelve Days of Problems

Only two weeks since my last AI post, but a lot has happened – and it isn’t good.

I don’t deal on speculation and rumors any more than future-only “progress,” so you won’t read here about the growing set of people thinking that Nvidia stock is heading for a steep decline – as with AI’s projected, expected, touted world-beating capabilities and applications, we’ll deal with it when it happens.  But here are harder concerns.

In “The age of AI BS” (Business Insider, March 27^th), Emily Stewart told us about “”AI washing,” or companies giving off a false impression that they’re using AI so they can amp up investors,” which has precipitated formal charges and legal settlements, as well as ChatGPT’s marketing campaign being mainly “to raise money, attract talent, and compete in the hypercompetitive tech industry,” a situation where “overselling has become a near-constant of the AI landscape,” and “it’s not super clear what the present capabilities of the technology even are, let alone what they might be in the future, so making bold, concrete claims about the way it’s going to affect society seems presumptuous.”  A “senior industry analyst” said that “a lot of these companies are not yet showcasing exactly what type of revenue they’re getting from AI yet because it’s still so small.” Stewart ended with “anyone who tells you they know exactly what is going on in AI and where it’s headed is lying.”

In response to a good and timely question, Zvi Mowshowitz informed us, in the New York Times on March 28^th, “How A.I. Chatbots Become Political.”  Although “our A.I. systems are still largely inscrutable black boxes” according to political-view assessment tests, most lean liberal and to some extent libertarian.  The biases may have been introduced during “fine tuning,” when technicians adjust outcomes, and from earlier developments, “because models learn from correlations and biases in training data, overrepresenting the statistically most likely results.”  There are now three versions of one major product, LeftWingGPT and RightWingGPT, which were evaluated as matching their names, and DepolarizingGPT, which still skewed slightly to the libertarian left.  A cause for concern, as “we may have individually customized A.I.’s telling us what we want to hear,” which may not end up being constructive.  This area, still, looks under control, but all should be aware of the biases these tools all carry.

Could it be, already, that “A.I.-Generated Garbage Is Polluting Our Culture” (Erik Hoel, The New York Times, March 31^st)?  In places, anyway.  One the author documented is recent academic peer reviews, which showed expressions known to be “the favorite buzzwords of modern large language models like ChatGPT” 10 to 34 times as often as in 2022.  Another example is nonsensical and uncorrected videos for children.  That sort of thing could cause a state “when future A.I.’s are trained, the previous A.I. output will leak into the training set, leading to a future of copies of copies of copies, as content (becomes) more stereotypical and predictable,” called “model collapse.”

Moving on, “It looks like it could be the end of the AI hype cycle” (Hasan Chowdhury, Business Insider, April 3^rd).  Not yet, though with the amount of accumulating doubt that may happen soon.  Along with sky-high importance assessments from Bill Gates and Elon Musk, Gary Marcus, who testified to the Senate on the technology, “predicted the generative AI bubble could burst within the next 12 months,” noting that “the industry is spending much more money than it’s raking in,” a great deal given that, per Crunchbase, “generative AI and AI-related startups raised almost $50 billion last year.”  Overall, “the verdict is still out on whether the companies behind foundation AI models dependent on expensive chips can turn their products into viable, profitable businesses.”

Back to a problem mentioned above, “Big Tech needs to get creative as it runs out of data to train its AI models.  Here are some of its wildest solutions” (Lakshmi Varanasi, Business Insider, April 7^th).  If, “according to Epoch, an AI research institute,” less than three years from now “all the high-quality data could be exhausted.”  Companies will need to do something, which could include “tapping consumer data available in Google Docs, Sheets, and Slides” which Google has already contemplated; buying an entire large publisher, such as Simon & Schuster, for its copyrighted information; “generating synthetic data,” created by AI modules themselves, using speech recognition to tap YouTube videos; and incorporating pictures from Photobucket, which hosted those from former large social media sites Myspace and Friendster.  In the meantime, we saw “How Tech Giants Cut Corners to Harvest Data for A.I.” (Cade Metz et al., The New York Times, April 6^th).  One way was the same YouTube idea, which was input into GPT-4 – this piece also mentioned the solutions Varanasi revealed. 

On the positive AI side, all I have seen these past two weeks is incomplete and future-bound.  For its actual advancement, they were a loser for artificial intelligence.  Will it get better, or worse?  I will keep you informed.

This Morning’s Report: Good Jobs News for Almost Everyone

 

The Bureau of Labor Statistics Employment Situation Summary came out of embargo 17 minutes before I wrote this sentence.  From a national economic perspective It was everything we could have hoped for.

Net new nonfarm payroll employment once again blew away its published estimates, reaching 303,000 instead of 200,000.  Seasonally adjusted and unadjusted unemployment fell 0.1% and 0.3% respectively, to get to 3.8% and 3.9%.  The total jobless count was off 100,000 to 6.4 million, of which those out for 12 months or longer remained at 1,200,000.  The unadjusted number of people working increased over a million, to 161,356,000.  There were 94,814,000 claiming no interest in a job, down 66,000.  Those working part-time for economic reasons, or keeping such positions while looking unsuccessfully for full-time ones, lost 100,000 to 4.3 million.  The two measures of how common it is for people to be in jobs or one step away, the employment-population ratio and the labor force participation rate, each improved 0.2%, realizing 60.3% and 62.7%.  Average private nonfarm payroll hourly wages rose more than inflation, 12 cents per hour, to $34.69. 

The American Joh Shortage Number or AJSN reversed course from last month, falling over 700,000 as follows:

The largest change sources this month were the unemployed, people not looking for a year or more, and those discouraged, contributing 329,000, 264,000, and 125,000.  Compared with a year before, the AJSN gained 462,000, more than that from higher official joblessness.  The share of the AJSN from that was 36.4%, down 0.4% from February.

Except for the second-to-last line, do you see anything bad about the above?  I don’t, except that some will focus on the lowered probability of interest rate cuts soon.  That is understandable, but let us hope that stocks will be more affected by the formidable strength and robustness of our economy.  These additional workers will spend.  Again we added vastly more jobs than our population increase could absorb, added people to the labor force, continued pay increases at a clearly justified level, and reduced the count of those wanting to upgrade to full-time.  Unemployment is still higher than it was a year ago, per the report “in a narrow range of 3.7 percent to 3.9 percent since August 2023,” and the 1.2 million still looking after a year or more could improve.  But those are distractions, and at least interest rates, where they are, appear fully justified.  The turtle stretched and took a large step forward. 

Artificial Intelligence Today: Combating Problems, and Drawing on the Morrow

As market capitalization surges and the press tells us about more and more noncurrent things, what is happening with AI now?  Mostly it is causing trouble. 

In “The AI Culture Wars Are Just Getting Started” (Will Knight, Wired, February 29^th), we got a recap on and views of difficulties in the front of the news several weeks ago.  As I documented in my last AI post, Google and Alphabet’s products were showing white male historical figures as female and darker-skinned.  Both CEOs apologized, Alphabet’s internally and Google’s publicly, and Knight suggested one mishap was because “Google rushed Gemini,” but a Mozilla fellow held “that efforts to fix bias in AI systems have tended to be Band-Aids rather than deep systemic solutions,” which makes sense, as fully retraining the products would require great masses of data and money.  Whether either Google or Alphabet were trying to make ideological statements is irrelevant now – they have huge business problems many customers will not tolerate.

“Silicon dreamin’,” in the March 2^nd Economist, was about the situation where “AI models make stuff up,” and asked “Can researchers limit hallucinations without curtailing AI’s power?.”  Such responses can be “confident, coherent, and just plain wrong,” so “cannot be trusted.”  Unfortunately, “the same abilities that allow models to hallucinate are also what make them so useful,” and “fine-tuning” them may make nonsense answers more common.  Overall, “any successful real-world deployment of these models will probably require training humans how to use and view AI models as much as it will require training the models themselves.”

Another way artificial intelligence may be making enemies come forward is in “A New Surge in Power Use Is Threatening U.S. Climate Goals” (Brad Plumer and Nadja Popovich, The New York Times, March 14^th).  With subtitle “a boom in data centers and factories is straining electric grids and propping up fossil fuels,” the piece documented that demand for electricity grew steeply from 1989 to 2007, was mainly level between 2007 and 2022, but is now forecast by companies selling it to trend upwards through 2033.  The main causes are electric vehicles, more manufacturing, and “a frenzied expansion of data centers” with “the rise of artificial intelligence… poised to accelerate that trend.”  While renewable energy sources can cover much of the increase, they cannot reasonably be expected to cover all of it.

On the user side, we were asked to “Keep these tips in mind to avoid being duped by AI-generated deepfakes” (Fox News, March 21^st).  Such things of ever higher quality have been in the news for months, and “video and image generators like DALL-E, Midjourney and OpenAI’s Sora make it easy for people without any technical skills to create deepfakes – just type a request and the system spits it out.”  Such things “may seem harmless,” “but they can be used to carry out scams and identity theft or propaganda and election manipulation.”  Fewer of these now have “obvious errors, like hands with six fingers or eyeglasses that have differently shaped lenses” or “unnatural blinking patterns among people in deepfake videos.”  Advice here was “looking closely at the edges of the face” for skin tone inconsistent with elsewhere on the person, “lip movements” not agreeing with audio, unrealistic teeth, and “context,” showing “a public figure doing something that seems “exaggerated, unrealistic or not in character,” or not corroborated by “legitimate sources.”  Generally, we need to be more skeptical about photos and video clips of all kinds – that will happen remarkably quickly.  In the meantime, though, “Fretting About Election-Year Deep Fakes, States Roll Out New Rules for A.I. Content” (Neil Vigdor, The New York Times, March 26^th), such as advertisers being required to disclose AI-generated content.  There are now, per Voting Rights Lab, “over 100 bills in 40 state legislatures.”

Beyond these and other troubles, how much AI is out there in corporate settings?  In “Meet your new copilot” (The Economist, March 2^nd), we learned that, despite massive company valuations, “big tech’s sales of AI software remain small.”  Specifically, “in the past year AI has accounted for only about a fifth of the growth in revenues at Azure, Microsoft’s cloud-computing division, and related services.”  As well, “Alphabet and Amazon do not reveal their AI-related sales, but analysts suspect they are lower than those of Microsoft.”  Two different consulting firms were not taken with current usage, as one said, “it will take at least two years to “move beyond the hype,” and another determined that “adoption of AI “has not necessarily translated into higher levels of productivity – yet.””

Per Oxford Languages, “morrow” is an “archaic” word meaning tomorrow, “the time following an event,” or “the near future.”  I have seen “drawing on the morrow” used as recently as the last half-century to mean getting a questionably justified advance.  I propose we use it that way for artificial intelligence progress.  AI may turn out to be a gigantically important business product, but it isn’t yet.  No matter how much its producers’ stock is worth, we can’t give AI credit for what it may or may not accomplish someday.  That would be drawing on the morrow.  That’s not how we want to deal with reality.