Artificial Intelligence is Pushing Against Its Boundaries

Over the past several months, articles chronicling the state of AI had something in common.  What is it?

The first, by Nico Grant and Cade Metz in the June 12^th New York Times, “Google Sidelines Engineer Who claims Its A.I. is Sentient,” described a sad situation where a worker’s views, though milder than those of many in the field of consciousness, got him placed on leave and, after the article was published, fired.  Some eminent observers in that field think that consciousness, about which we know very little, may come from computations, making it inherent even to $5 pocket calculators.  Blake Lemoine, the engineer, could be right, and none of us know that he isn’t.

Yes, now as always “We Need to Talk About How Good A.I. Is Getting” (Kevin Roose, The New York Times, August 24^th).  The author ran down recent artificial intelligence achievements, from generation of art of sorts based on requests giving its subject, to one able to “predict the three-dimensional structures of proteins from their one-dimensional amino acid sequences,” and in the process solving “what’s known as the “protein-folding problem,” which had vexed molecular biologists for decades,” and recently making “predictions for nearly all of the 200 million proteins known to exist – producing a treasure trove of data that will help medical researchers develop new drugs and vaccines for years to come.”  Roose claimed that “the conversation in Silicon Valley is starting to shift,” as experts “now believe that major changes are right around the corner, for better or worse.”  He said “regulators and politicians need to get up to speed,” AI companies should communicate better about what work they are doing, and “news media” must improve at explaining it – all constructive ideas.

The sentience would be helpful to fulfill “One Man’s Dream of Fusing A.I. With Common Sense,” by Steve Lohr in the August 28^th Times.  His startup, Elemental Cognition, is working to develop artificial reasoning along with the pattern recognition AI systems have excelled at.  This company trains them through “machine learning algorithms” to change human-language document contents into “a form a computer can interpret.”  A strong potential growth area, with success promising but uncertain.

I have been predicting that foreign workers would greatly reduce both jobs and pay for American computer programmers, but will automation instead be their demise?  As shown in “Coding Made AI – Now, How Will AI Unmake Coding?” (Craig S. Smith, IEEE Spectrum, September 19^th), it seems quite possible.  The author held that while programming and software development “appears poised to remain a very human endeavor for the foreseeable future,” at the same time “coding as we know it may indeed be doomed.”  It’s easy to imagine, within a decade or two, giving human-language written instructions to an AI system and asking for it to create a program, letting the facility use its own chosen methods, not necessarily computer languages as we know them, to do that, and consistently getting excellent results.  As AI develops, “hand-coding software programs will increasingly be like hand-knitting sweaters.”  If that happens, it would be impossible not to expect the number of positions to vastly decline.

On the applied section of artificial intelligence known as robotics, we have reached a historic point, as a certain “Robot Fast Food Cook Costs Less Than Half a Human Worker” (John Koestler, Yahoo Finance, September 28^th).  Now available to rent for about $3,000 per month, The Wingman might be the first restaurant robot fundamentally cheaper than employees.  If it succeeds under sustained production pressure, it could be used by tens of thousands of eateries nationwide, replacing workers in its wake – and every effort to mandate or just implement higher wages will only speed the transition.

This concern is described further in “Nouriel Roubini:  Why AI poses a threat to millions of workers” (Daniel Howley, Yahoo Finance, October 18^th).  The interviewed author reminded us that people in fields from not only the obvious algorithmic ones but, as above, in the arts, will not only be thrown out of work but will have shaky prospects elsewhere.  Nothing’s new here but the scope, and indeed, if I need to “create a script or a movie, or make a poem, or write… or paint, or even (write) a piece of music” that does not need verve, I may soon not need human creators at all.  That driverless cars have stalled does not mean that all automated interfaces will do the same.

So what ties these six pieces together?  All are about artificial intelligence’s forward progress.  They combine for a huge, underreported news story.  AI is moving ahead in real life – we need to prepare.