Troubles with AI in Newspapers

I was chatting with Joyce J. at a meeting of a club Martha and I belong to. Joyce was a teacher.  She is now retired and dedicates her time to volunteer work.  I was describing what SCS does for newspapers.  As she listened she seemed to become more and more concerned.  

"What are you going to do about all the jobs your systems are eliminating?"  she asked.

I was surprised.   "Like what?"

"Are you going to help the people - maybe get them some training?"

"Why or how would I do that?" I said, somewhat taken aback.  "They are not my responsibility."

"You should be doing something,"  she continued.

"How could I afford  that?"

"Well, you could apply for a grant or something like that,"  she said.

Apply for a grant?  Now I was really mystified. The last thing I want to get involved in is applying for grants.  It's a lot of work and I've been turned down more often than not by both government and private organizations.  Besides, our newspaper customers are in 18 countries around the world.  This seemed plain silly.  Still her comments weighed on me all Christmas week.

I consider our company to be quite virtuous.  We hire interns and pay them.  We make sure they have really useful experiences with interesting work, fair wages and hugely valuable mentoring.  Many report our internships are one of the most rewarding experiences of their education. (I get the pleasure of telling old stories to new, young faces.)

We build trusted newspaper systems that are both extremely reliable and affordable.

We employ great talent - super people that our customers often say are the best they have ever worked with.  We don't outsource.  We are a Made in America Company in the best sense of the phrase.

We help newspapers when times are tough.   We have saved them when there were disasters, both natural and terrorist made.

Martha founded SCS in 1976 and since 1983 we have run it together.   We consider ourselves model business people.

Still Joyce's comments caused me to lose sleep over the next week.

Sure, newspapers use our systems to reduce labor costs.   Many have done staff reductions with generosity.  Others - not so much.

How do our applications like Layout-8000 and SCS/ClassPag save labor?  They automate display ad dummying and classified pagination. They come with middleware that ties disparate systems together.  They are industry standards.  Yes, these days there are fewer jobs in newspapers. Part is due to the realities of being a print medium in a digital advertising age.  Even so, we have literally saved at least a half-dozen newspapers from going out of business over the last 36 years. 

I kept thinking, "We are the good guys."

Wait, sometimes when we are cast as bad guys, the other side fights us.   We call this PICNIC.  Problem in chair, not in computer.  With every new labor saving technology Luddites fight the inevitable.   Sometimes they even win, but the win doesn't last long.  It seems that it is when you provide software that replaces jobs that involve problem solving and creativity that trouble is most likely to arise.  Bright people may think that it is all right for machines to replace John Henry, but not them.  Their work is brain work, not muscle work.  It takes smarts to dummy a newspaper.  It takes smarts to paginate classifieds.  The more ads, the more difficult.   The truth is that making smart machines has been my life's work.

There was a time not so long ago that chess programming was the drosophila melanogaster of computer science.  (Drosophila is the fruit fly geneticists use in experiments.)  Artificial intelligence (AI) was forever going to be the next big thing.  Before I got my BS in computer science, I had been Delaware's state Chess Champion for three years.  Creating a chess machine became my passion.  I learned everything about AI programming I could.  Writing such programs is really tricky and fun, since for the most part, they must just work right from the start because the computations are so large you can't hand simulate their logic.

My computer science research focused on finding the simplest algorithms that improved chess play. I had several articles published reporting my results.  This remains a significant theme for AI research today. There have been major breakthroughs very recently.

In 1973 I brought this knowledge to the American Newspaper Publishers Association Research Institute (ANPA/RI), the trade association that has had a number of  names since then. I thought to myself, "You are going to get to apply your computer science knowledge and skills to serving the free press.  You will help newspapers hold the powerful accountable.  How cool is that!"

I completed my first and second ANPA/RI assignments: 1) making DEC's RSX11-D operating system run on the lab's anemic PDP 11/45 computer and 2) then writing the software and logic for what was called the Laser Page Setter. It was perhaps the first technology (circa 1974) to digitally image newspaper pages.  I was then asked to figure out what the lab's computer applications department should do next.  I'd been made the department's manager by then. I also upgraded the staff with my former Lehigh University students  I had taught as adjunct faculty.

The ANPA had supplied significant funding to MIT for developing newspaper design and pagination technology.   I offered that our group could produce practical, deliverable solutions for ad dummying etc. based on MIT's research but without their help.   And so we did.  Unfortunately, we didn't get the left-over MIT funding ourselves. When it came time for the ANPA to move to the Washington area, we had over three dozen operational sites for Layout-80, as it was called then.   The folks at the ANPA were very kind to Martha and me, perhaps because I did as told and sold licenses not just to Martha's SCS, but five other established newspaper vendors as well.

ANPA's chairman of the board, Bill Marcel publisher of the Fargo Forum, insisted on becoming SCS's first customer for Layout. Having always been a rebel, it was nice to feel the love.

So what about Joyce.   I surely wasn't feeling the love from her.  We met again last Sunday.   I had been so disturbed by her comments, I wanted to have another go at seeing her point of view.   Maybe I could justify thinking of myself as one of the good guys again. 

Think about it.   There are people and tasks.   You could focus labor savings from either direction.   Eliminate jobs or tasks. The troubling issue is that having a person do a job that a machine can do as well or better, dehumanizes the human.  It is also dehumanizing to think of yourself as an item on a spreadsheet for improving ROI.  There has to be a better way.   Bright people are adaptable. Those who dummy advertising and paginate classifieds are, of necessity, quite bright.  There has to be a better way. 

When newspapers acquire our systems, they should look to eliminating tasks with our AI and find ways to redeploy the people.

Such a recommendation is, perhaps, enlightened self-interest.   Sometimes purchases of our systems are delayed. Management might just be retaining valued staff by keeping the manual work that they do.   Of course, this only works for a while. When savings are critically needed, change comes bluntly, in a hurry.

Perhaps this is a better way.  Retrain and redeploy.

Joyce liked the idea that she had such an impact on me.   She even seemed to like my new way of looking at deploying AI systems.  We thanked each other for sharing.

P.S.
Multi-billionaire, artificial intelligence expert Kai-Fu Lee said that he believes 40% of the world's current jobs will be done by AI machines in 15 years.  He claims that Chinese companies like his, and not US Silicon Valley companies, will be the principal creators and builders of these machines.