You are being watched...

I mentioned in a recent blog post how Artificial Intelligence (AI) may intrude into the practice of commercial image making, likely eliminating the role occupied up to now by photographers. Today I came across an illustrated discussion about how AI might take on the role of the art critic. The centerpiece of the discussion was quite a fine photograph by a fellow I follow on Flickr, Leon Van Kemenade.

What Leon did was to request an AI critique of his photograph of a woman on the street holding an umbrella. The response was appropriate, but lacking in some interesting ways that the photographer describes quite well. The thrust of his argument was that the AI critic made a reasonable but inadequate response which omitted a consideration of how the photographic image might be interpreted.

Several things occurred to me when I considered the photograph and the discussion, starting with the idea that AI is going to intrude on every human endeavor including those activities which may have until now been considered uniquely human. I agreed with Leon's analysis of the inadequacy of the AI critic's evaluation of his photograph.

I also think that the shortcomings of the performance might also be considered due in part to the fact that the prompt was inadequately framed. It seems highly likely that adding some detail as to what was required would have produced something closer to what was wanted.  Also, making some more information available to the AI critic would have produced a learning opportunity which would certainly have altered future responses to similar questions.

Another angle on consideration of the AI response was that there was some value in spite of of the perceived inadequacy.  For instance, it is not hard to see that an AI summary of the issues in question could prove to be a useful time saver. In fact, students at all levels are using the technology in precisely that way now.

Many are understandably nervous and critical about the perceived threats from widespread AI implementation.  However, it is probably also important not to overlook potential benefits. An example of this balancing act is to consider one AI implementation that has now been in place for a considerable period of time -- the application of AI to the game of Chess.  I.E.

GPT-4o mini (via DuckDuckgo):

"A computer first beat a chess grandmaster on May 11, 1997, when IBM's Deep Blue defeated the reigning world champion, Garry Kasparov, in a six-game match. Deep Blue won the match with a score of 2-1, making it a historic moment in artificial intelligence and computer science. This event marked a significant milestone, demonstrating the capabilities of computers in strategic thinking and complex problem solving."

That development seemed to foretell the possible coming supremacy of machines over humans.  That may turn out to be true, but in the meantime AI programming has provided a means for practitioners of the art of Chess to deeply investigate potential strategies.  It also seems very likely that the opportunity AI provides for online chess competition has contributed to the appearance of a great number of youthful chess prodigies from all over the world.

So, it seems important not to jump to conclusions about such a complex issue. Given the billions that are currently being poured into AI development there is likely no way to stop the train.  Better, perhaps, to consider to what ends the technology will be applied and to whose benefit.

Language Proficiency; Making the Right Moves

I have completed the second week of the Coursera on line linguistics course.  I am pleased to say that the course has already provided me with some answers to questions I have had about language acquisition and proficiency.  Of course, there also have been answers to questions which had not previously occurred to me.  The exercises in the fundamentals of phonetics, for instance, gave me some insights into how consonant and vowel formation actually takes place and how it can be systematically analysed.

In the end of the week wrap-up video the professor, Marc van Oostendorp, gave a glimpse into the course's demographics.  About 42,000 participants from 190 countries are now engaged in studying the basics of linguistics.  The email referring to participation mentioned that it would be revealed in the video what was the age of the youngest and oldest students.  I approached that revelation with some trepidation as there was a possibility that I might be the oldest.  In fact, while I am certainly located in the narrow right tail of the age distribution curve, the honor for oldest went to "Margaret", an eighty-year-old South African.  The youngest course participant turned out to be just fourteen years old and a resident of India.  It would be interesting to see more on the statistical profile of the student body; I'm sure the U. of Leiden will be paying close attention to those numbers.

Participation in the linguistics course has provided me with the opportunity to reflect on my own struggle to attain some proficiency in Spanish as a second language.  On the one hand, it is a little dispiriting to observe that a life-time of study and practice has not resulted in anything close to perfection.  At the same time, I am encouraged by the fact that I seem to have made some progress over the past couple of years.  I am reading more challenging books than I was comfortable with previously and I feel that my comprehension has improved.  It seems unlikely given the state of my memory that any improvement is due to vocabulary enhancement.  Rather, I would attribute my perceived gains more to some improvement in attention and engagement.

I have found that making a conscious effort to focus on the meaning of the text I am reading actually pays off.  I tested this idea recently by going to the UNM library and sitting down to read a short story by a writer who was new to me, Mercè Rodoreda.  While I did not have the crutch of the Kindle translation dictionary and did not consult a dictionary in book form, I was able to follow the story line without great difficulty just by taking my time to absorb the writer's intent.  Given those encouraging results I went on line and ordered a copy of the author's best-known novel, La Plaza Del Diamante.

At the other end of the spectrum from me is that fourteen-year-old Indian kid.  He is taking a college-level course taught in English, just one of several languages he speaks.  Those facts highlight the well-known capacity to absorb languages at an early age, which seems to atrophy as we get older.  Receptivity to language acquisition is likely based in part on pattern recognition capacity, and that may well extend to other areas of learning as well.  For instance, I am thinking of a news item I saw recently about the chess grandmaster, Sam Sevian, who attained his title at the age of thirteen.  As seems to have been the case with many other chess greats, Sevian was introduced to the game by his father before he was old enough to start school.  There may be a genetic component to Sevin's success in chess given his father's expertise in the game, but it also seems possible that missing that early opportunity to be exposed to the game could have closed the window on achieving the extraordinary potential which was initially available.

the six pieces

Chess pieces are a nice subject for modeling practice. They are made mostly by spinning a half profile. The knight must be carved or formed by other means. SketchUp has a limited capacity for such functions, which in part explains the rather blocky look here. I may have another go at this subject once I have attained more skill with Blender.

errata:
Handling my chess set prompted me to pick up an old chess primer by Fred Reinfeld. I was reminded that there are - strictly speaking - only five chess "pieces": the King, the Queen, the Rook, the Bishop and the Knight. As Reinfeld notes, "the Pawn is just a pawn".

SF Queens

In 1971 Bobby Fischer was on the verge of becoming the World chess champion. In the U.S. interest in chess was at an all-time high and games could be found in progress at every lunch break. I had finally landed a job as an eligibility worker in San Francisco's Food Stamp program. Getting my first paycheck, I went next door to a Goodwill shop and bought an antique bone chess set for ten dollars.

I was thrilled with my purchase of the chess pieces because of their graceful design. Over the years, I have searched for clues about the origins and age of my chess set, but to no avail. The pieces are made from columns and disks of bone, turned on a lathe and threaded with tap and die for assembly. Red and white were the standard colors for chess pieces around the turn of the Century, and whale or camel bone often provided the raw material. My set exhibits the basic design simplicity that was introduced by Staunton in the late 19th Century, but the lines and proportions have a wonderful combination of strength and delicacy which I have never seen in another set.

The image above and a few others are available on the Free Wallpaper Downloads page of my web site.

Update:
I did a Google image search on "bone chess pieces" this morning and turned up an image of a set nearly identical to mine in an antiques sale listing. The context implies that the set is of American origin, though there is a good chance it is English. The red paint in the advertised set has faded, possibly from being displayed in a location too exposed to the light.



"[Item] 109, Ivory or bone chess set. Early 20th century (or earlier). [Estimated value:] $50-$150."

And then I found another image of the set on Flickr. I also found a couple ebay listings which describe the pieces as being "antique ox bone". I would still very much like to know who designed and produced the set, and when.

The chess set turned up again in an Aug. 29, 2014 photo essay in the New York Times.  This time it was set up ready for a match in front of a New York psychiatrist.

game pieces

One of the most interesting features available in all 3D computer graphics programs is the capacity to emulate forms produced by a lathe. In SketchUp the "follow-me" function is most often used for this purpose. A 2D cross-section of a form is drawn, and that is then spun around a central axis to produce the rounded 3D form. While one can thus produce round vase-like objects, the process is also useful for creating new shapes and surfaces through the intersection and joining with other shapes.

The pawn figures were actually an intermediate step in the creation of a thumb-latch in the model I am working on at present. After spinning the form, I sliced it vertically through the middle, and then laid it over on the newly-created flat surface. I like the simple elegance of the unaltered form, however, and perhaps I'll use it as-is in some future composition.