Dopamine And AI, Part 2

October 16th, 2024 No comments

So we’ve postulated that a Dopamine response can be targeted by businesses to increase sales or to shape behaviors.  Employing sounds, smells and sensations, are known actionable triggers for consumers.

But let’s pair that with rapidly evolving Artificial Intelligence tools.  Mostly everyone in the developed world has exposure to streaming music and entertainment platforms.  When they were first introduced, they were simply tools of convenience as mobile phones once were.  But with the deployment of their internal algorithms, these services now create suggested material based on your viewing or listening proclivities.  Thus, being fully attuned to the material that gives you the most satisfaction, which means Dopamine hits, they continue to feed you material based on their predictive models. You are now a commodity to whom marketing will be specifically tailored.

No doubt, they’re succeeding with these tools.  However, one of the shortcomings of AI (at least thus far) is that projections are necessarily based on past events.  They use pattern recognition and extrapolate trends based on previous samples.  One could say that it’s as if you are driving forward using the rear view mirror.  If your playlist is full of Barry Manilow, it’s unlikely that anything by The Beastie Boys winds up in your suggested rotation.  If you’ve watched more than one Reese Witherspoon movie, you will have tons of blonde ingénue films populating your suggestion box.  In this instance, use of AI is similar to politicians who tune their message to whatever they think people want to hear.

Which is fine.  Except that this does not provide something that humans innately need….the desire for variety and novelty.  While the old and familiar are stable bets, humans generally have a need for variety and novelty rooted in natural curiosity.  This is why human input cannot be completely replaced by the emerging creep of AI in our lives.  People should be aware that present popular AI platforms such as Chat GPT and Grok rely mainly on their ability to scour and assemble vast known data sources to create an appropriate response to a defined problem.

AI’s capability for pattern recognition is unmatched.  These platforms can very rapidly give solutions to existing problems based on the vast database of similar events that are known. As of right now, they are not yet able to solve problems for which there are no known solutions. Of course, this may change with time, but as an example, in the medical field, they can’t provide the cure for cancer, BUT they will sift through mountains of known data to provide what has been effective therapies for resolved medical issues.  AI won’t be able to solve conflict issues in the Middle East nor be able to predict the Super Bowl winner; it can only give statistical probabilities.   AI is not a magic 8 ball.

I can see these tools being used to drastically reduce the backlogs in the legal and justice system.  A large part of the law is based on historical legal precedence, thus running the variables through an AI filter could eliminate the need for parsing words, nonsense arguments and activist judges.  We can certainly see the use of AI in the field of medicine for treating most of the routine ailments that afflict people.  Even a casual dive into the dozens and dozens of AI content creation sites is astounding and it’s all just beginning.

While AI is probably the greatest technological leap by man post the industrial age next to microchips, it also brings with it some complacencies and dangers as yet unknown.  While it can figuratively assemble all of the available ingredients in a pantry to make a meal, it cannot create unique solutions outside of the known universe of recipes.  At the very root of any data set are human inputs which then result in acceptable outcomes

For the sake of illustration, imagine that we are in the 1800’s and leeches are commonly employed in the cure of many medical issues.  If AI were available at that time, any search for a treatment for circulatory issues would result in the application of leeches….because that was the known and acceptable treatment of that era.

In today’s world, it’s very possible that requesting an answer to a question such as “who would make the best president?” may return an answer of Kamala Harris, depending on whom entered the variables.   Thus, while AI can satisfy specific requests with logical outcomes, the baseline of data points had to have been created by human input.  We all know the programming mantra of GIGO, garbage in, garbage out.  An apt illustration of this is the Wikipedia site.  While many refer to this site for research, many are not aware that entries are edited for ‘appropriateness’.  Thus, if this site was referenced as part of the process for determining an answer to a question, the response will be skewed because of the flawed input.

With broader employment of AI tools in every aspect of our lives, there is the real danger that it can assume the role that television once did when it became a ubiquitous household appliance; that is, people will think something is true because they saw it on TV.  Even at this early stage, it’s often difficult to discern reality from deep fakes in many AI video presentations. As Abe Lincoln purportedly said, you can’t trust everything you see on the internet.  Knowing what we know about human behavior, it can be very easy to create narratives that reinforce acceptable worldviews just as they play algorithmically created music on a Spotify feed.   You will hear what the machine thinks that you want to hear and you will experience what the machine calculates what you want to experience.

Notwithstanding the enormous power and potential of AI to modify our lives, the key task of real innovation and imagination will always require the human element.  The source code will always have at its root, human input. Ironically, the abilities of AI may make populations more complacent, more malleable and less innovative and more derivative. This is a real danger.  While certainly we are exposed to new things that seem to appear daily, many of those things are derivative or just iterations of an existing thing. In fact many of these new ‘things’ have a certain sameness to them.  Think Gothic dragon films, rap music or SUV’s.

But no amount of AI writing tools will replicate the creative mind of a J.K. Rowling nor the insightful, sensitive humor of a Bill Watterson.  No AI tool will replace the stark depiction of the human condition like a Solzhenitsyn.   No amount of data crunching will replace the mind of an Elon Musk in pushing the boundaries of technology and imagination.  People like that will always be at the leading edge of human aspiration, not content to listen to the usual comfortable playlist that life has made for them.  It turns out that achieving satisfactory, comfortable results is not the end game for bright people, nor should it be for mankind. Innovative pioneers will continue to make their own original life playlists with great unpredictability.  This brings to mind the Roadrunner cartoons.  Even as Wile E Coyote brings sophisticated technology to use, the Roadrunner always finds a way to escape.

Dopamine And AI, Part 1

October 15th, 2024 No comments

People who read my pieces know that an underlying theme of many of my articles concerns human behavior; what influences people and how, especially in large groups.  Apart from direct physical coercion, it turns out that the things that people choose to do out of their own free will has much to do with a hormone and a type of neurotransmitter called Dopamine.  It’s explained as a sort of chemical messenger, made in our brains and according to WebMD…

“… It plays a large role in determining how we feel towards pleasure and rewards.   It’s a big part of our unique human ability to think and plan.  It helps us focus, work towards goals and find things interesting…”

So, among other things, Dopamine determines what makes things pleasurable for us.  Some basic examples would be the smell of morning coffee or a bakery, or the smell of a baby to their parents.  It could be a warm towel after a bath, or the throaty sound of a sports car or a favorite piece of music.  Even the anticipation of these things will trigger pleasurable responses in people.   This knowledge has been harnessed for years by corporations marketing any and all sorts of things to the public.  By being able to access people’s Dopamine triggers, people will feel the need to purchase their products in order to get their hit of Dopamine and thus, pleasure.  In a 2003 study by Paul Phillips and some colleagues at the University of North Carolina, they found that rats would experience a spike of Dopamine when they were only anticipating hitting a lever that would give them a jolt of pleasure.

This study led to many other researchers giving insights into the role of Dopamine and addictive behavior in humans. The commercial harnessing of this hormonal feature has been key in the success of many successful consumer companies in the past decade.  Think about people’s recent generational addiction to their phones. Think about the joy of hitting the buy button for an online purchase and once again when the Amazon or UPS truck pulls up. The actual item purchased is irrelevant since very shortly, the consumer will be back at it, looking for the next ‘reward’. Think about addictions to gaming, to online porn, to gambling.  In all of these cases, the anticipation of the event triggers Dopamine surges to the brain.  One of the best exploitations of this weakness in people are lottery tickets.  Very few people actually win the big prize, but small inconsequential ‘wins’ are enough to encourage people to buy them again and again, the same formula as in casinos.  Even though they’re unlikely to win, reports of someone winning a big prize is enough to trigger even more buying.

The rise of the popularity of day trading is very much related to the Dopamine hits received by day traders collecting numerous small wins even while losing net in the long term.  The action is the drug.

This phenomenon is also employed with something as benign as music.  It’s not a coincidence that in many retail businesses that encourage shopping, certain targeted music is played to increase the Dopamine hit for shoppers.  If you’re in an upbeat mood, presumably, you’re more inclined to spend. By late November, most retail stores will have Mariah Carey on a loop singing “All I want for Christmas”, although in this case, it may actually be counter-productive.

This knowledge of human behavior leads us into the more focused use of Dopamine effects….pairing with Artificial Intelligence.  I will discuss this further in the upcoming part 2 of this theme and of its relevance to Artificial Intelligence.