There is much fuss going on about AI. What do you think?

When I was a student in college and was introduced to Carl Jung in a Psychology course, Jung’s theory of a “collective unconscious” seemed impossible to substantiate. At the time he proposed the theory in 1912, molecular biology was a far distance from where it is today. There was no way to develop empirical evidence one way or the other on the unconscious minds of billions of people.

         Recently, I heard a short discussion about ChatGPT in a Jordan Peterson podcast during which I experienced an “eureka” moment. Your grandmother had been pushing me about AI for weeks. Wall Street Journal articles attracted her attention to the issue of AI in our future. Elon Musk’s recent sounding of alarm over AI only intensified her interest. I had given it little thought except in the context of how certain stocks might be affected in the near term by developments in AI technology. Turns out your grandmother’s and Musk’s concern about AI is something else entirely based on what I heard in that Peterson podcast.

         My first silly thoughts before that podcast were that, if Ais were able to control man’s weapons of mass destruction, they might conclude logically that homo sapiens were the greatest danger to the future of the planet and thus dispose of us while preserving all the rest of Earth’s flora and fauna. So, no nuclear explosions. Perhaps poisons in French fries, pork won tons, kidney pies, red wine, and beef tacos.

                   Confronted with your grandmother’s continuous thrusting of articles about AI in my face, I read and quickly concluded that there was no foreseeable way that man was going to be able to construct a sentient AI. Machines could follow orders, but they could never “feel” the reaction in a human that their words might cause or be made to feel remorse for telling a four-year-old that there was no Santa Claus. I held that position until I heard Peterson’s discussion with the founder of ChatGPT. My current thoughts require some background.

         In 1961 (the year Jung died) and my first year at Harvard, the discovery of the DNA helix was first revealed and was an immediate sensation in classrooms. While science was not my path, I always was interested in a variety of scientific subjects. When I could finally afford it, I became a subscriber to Scientific American from the time I graduated from law school in 1968 until I started my law firm in 1980. Accordingly, I observed from a distance the geometric gains that were happening in molecular biology. In the late seventies I recall reading of mitochondrial DNA. Today, it is purportedly used to trace ancestral roots.

          Passed from the mother with long stretches of time before small mutations occurred, mitochondrial DNA was reported to have started with “Eve” who lived 150,000 to 200,000 years ago. Supposedly there were seven or eight main groups of the DNA formed by small mutations 70,000- 100,000 years ago each with 2-3 subgroups formed 20,000-30,000 years apart. I recall thinking when I learned of mitochondrial DNA that Jung might have jumped on this development in biology -- had it occurred in 1912 -- to explain his “collective unconscious.” After all, our brains decide how we feel about things and carry our memories, including pleasing happenings and anxiety-filled moments. Mitochondrial DNA seemed to be a natural for carrying our species’ memories from one generation to the next. For example, no one had to teach me that a growling dog on my paper route was not a good sign for me. Did I have “Eve” of 200,000 years ago to thank?

             More recently, I read a study that suggested in Alzheimer patients the mitochondrial DNA in the brain shows ten times the deterioration rate compared to nuclear DNA in the brain. That seemed even more evidence that memories carried in the mitochondrial DNA may explain Jung’s “collective unconscious”. The original strand supposedly originated 150,000-200,000 years ago. The M mutation occurred 70,000 years ago. The H and X mutations occurred 30,000 years ago. In other words, the mitochondrial DNA goes thousands of years unchanged in our species. The existence of a “collective unconscious” contained within our species’ molecular biology seems less hypothetical 100 years after Jung posited the idea.  

            During the Peterson podcast, they discussed the fact that, while our feelings may be chemically induced within our brain, we are required to use language to express those feelings. The only way for another human to know that we felt something was good was to put it into words. Having studied the faces of hundreds of jurors and even more potential witnesses, I disagree that language is the only way human’s express feelings. However, I cannot disagree that our language carries within it a map to our emotions. Eureka!

             AIs are programmed using our language. If within that language one can discern the array of emotions we feel, AIs are learning that emotional array as they learn our language. But can they learn to allow a child to win a chess match to seduce interest in the game. And to teach the game by starting to take advantage of the first mistake in the next game to show a move to avoid. Now that I mention it, I suppose the AI can be programmed to play perfect chess against an expert and “teach” chess to a youngster. I guess the AI would ask, “What is your level of expertise?” and select the appropriate program among its – must we be open to pronoun diversity when dealing with AIs? -- millions of programs.

                However, someone must program AIs at the outset. What if I was that programmer. AI would view Federal government activity beyond providing military defense, interstate highways and bridges, and collecting taxes as extremely limited. There would be a small FBI and a medium-sized State Department. Environmental and stock fraud protection would depend on private litigation to fight it out in courts with juries determining the outcome rather than Federal bureaucrats.

           What if a Democrat was that programmer. A radically different view would be programmed. Federal bureaucrats would be conceived as far better able to ensure that citizens behaved in the manner deemed appropriate. The idea of self-determination would be frowned upon. Even more horrible would be the idea of competition with winners and losers.

           AIs are being thought of as neutral, but the people who program them are not capable of neutrality. We are all filled with bias even if subconscious. The most important element of picking my jury was posing a question and watching their eyes as they responded. The words might be of some importance, but the eyes told me all I needed to know. It occurs to me that one group AIs cannot replace are trial attorneys.

           I decided to try ChatGPT as more direct research. I asked a simple question at first. Its quick response was it could not answer my question because it was only programmed through September 2021. I then asked if it was discerning human emotion by the language we use. It responded quickly that it did not have emotions and could only provide facts. Finally, I asked why the United States dropped an atomic bomb on Hiroshima. It responded within three seconds with a concise but reasonably accurate array of likely reasons but noted that it remained a subject of much investigation with various interpretations.

       I was impressed. However, in that short exploration I did not develop any intuition of whether ChatGPT was on a path that was likely to make it sentient someday. I intend to keep playing with it to see if it is any better than exploring Wikipedia or doing a deep dive into a Google or Bing search. I already know that it is much faster, but it provides no citations to allow me to investigate the quality of the data on which it is basing its answer. At least, Wikipedia provides citations for the source of almost every statement it makes. I then can investigate for myself before accepting a statement as fact.

          It took hours of research but when your grandmother was diagnosed with cancer I found and read scores of medical journals on the subject. Thereafter, the questions I had for the group of doctors treating her surprised them for someone without a medical degree. I am not sure ChatGPT would have provided a similar array of information. What I do believe is that it will be able to prepare the 250-word essay that Maeve and Eli (my grandson’s children) will be required to hand in when they take a high school course. Hopefully, they would at least review and edit it with their own insights before handing it in as their work. But it would save time for them to practice violin or basketball or golf or dancing – more important skills than learning to take purported facts, competing interpretations, and discern your own view of things?

        In the meantime, I will not ask Chat GPT who won the Bama-UT game in 2022. I can’t recall offhand. Can any of you? Okay, once in 15 years does not a trend start.

Post Views : 21