Intelligence
On November 30, 2022, OpenAI released to the public ChatGPT, a natural language chatbot. Trained on vast quantities of written documents, it possesses an eerie ability to render responses to requests that are posed to it. It is presumed that it will continue to learn as a result of these human interactions, expanding its capability to interpret information and produced seemingly unique content.
I had been reading articles about it for several weeks before I decided to put it to the test. I asked that it provide a one sentence summary of two long paragraphs, with multiple bullet points, that were developed as vision statements for a client. Almost instantaneously after I had pushed “enter,” it responded with a reasonably good response.
New technologies, it seems, often require some period of adjustment on the part of the humans intended to benefit from them. Applications utilizing artificial intelligence will likely be no different. Various media are publishing reports of concerns from educators that ChatGPT will be used by students to write essays. New York City schools responded to these concerns by banning the site from school computers and devices.
Many of us, perhaps unwittingly, are already interacting with AI programs when we access some commercial websites. Sometimes they are helpful in responding to our requests while at others they are not. My expectation is that they will continue to improve over time, in much that same way as is expected of ChatGPT and similar programs.
What does this mean for those of us involved in knowledge work? Is there real utility to AI applications or are they just performing very sophisticated parlor tricks?
Let me answer the last question first. I expect those of us in management occupations will find more and more uses for AI programs, providing incentives for their continued development. Consider a simple use case. I often have occasion in my work to prepare reports on various topics. If the document is lengthy, I will include an executive summary. A program such as ChatGPT could easily and quickly produce one for me. Although it might require tweaking, my efficiency would be increased if the initial draft were prepared through such a means.
In this respect, these AI programs are a natural extension of the autocomplete features of word processing programs. I have been using this feature of Microsoft Word as I prepare this article. It saves me time by offering suggestions for how to complete words or phrases. Other program features identify misspellings and recommend improvements to grammar.
As the sophistication of language AI applications improves, should I be fearful that all my functions as a local government manager will be taken over by them? Such a future seems unlikely. How I perform in such a role involves more than the use of language.
In 1983, Howard Gardner, a Harvard professor and psychologist, identified eight types of intelligence. One, linguistic intelligence, is the domain in which ChatGPT operates. It is also the form of intelligence that is the basis for the Turing test, a method for determining whether a computer could be said to be intelligent.
This test, at least in its original formulation, relies upon the isolation of a computer from a human. Communication would be done only through text. The human would not know in advance whether he or she is engaged in a conversation with another person or a computer. It seems likely, now or in the near future, that applications will routinely pass the Turing test.
Does that make them intelligent? Not if we insist that they display all eight forms of intelligence theorized by Professor Gardner. I can imagine that computers will be able to display forms of intelligence that mirror that of humans in areas other than linguistic intelligence and will be able to pass comparable Turing tests for them. I am doubtful, however, that they will ever have the sophistication to display interpersonal, or emotional, intelligence. It is in this area that I think the true value of those involved in management will be demonstrated.
In a past article, I focused on the nature and importance of insight, which I summarized as the ability to identify connections between pieces of information and, in such a way, develop new ideas. It might have been shortsighted on my part to overlook the potential for AI applications to do the same thing, maybe quicker and better. The information I was considering as I wrote the article, however, was largely devoid of individualized emotion.
We learn over the course of our lives how to interpret the signals given to us by fellow humans through our daily interactions with them. We recognize their moods and facial expressions and adapt how we deal with them based on that information. This is a highly sophisticated enterprise. Those of us who succeed, particularly in executive roles, are likely to be those most skilled at “reading the room.”
At any given time during my career as a city manager, it has been necessary for me to ascertain the motivations of governing bodies with as many as nine members and executive teams with as many as a dozen members. I have interactions with members of boards and commissions and those who lead other public and nonprofit organizations in the community. Citizens come to city hall with their complaints and concerns. My responses to these individuals must be adjusted so that they are appropriate to them as unique persons with histories specific only to them. There is no one-size-fits-all approach to interpersonal intelligence. More importantly, it is a skill that can be perfected only by practicing face-to-face.
It is this characteristic that represents the greatest challenge for any who would suggest that AI programs can develop interpersonal intelligence at the same level of proficiency as they will develop linguistic intelligence. Becoming highly skilled in the area of interpersonal intelligence is an iterative process achieved through interactions with another human.
Consider a game of Texas Holdem poker. There is information that is shared by all players: the rules of the game, the size of the stacks of chips held by players, their relative positions at the table, their history of play, the cards turned over in the center of the table, and so forth. There is also information held only by each individual player, specifically the two cards in the player’s hand. Because this is a game of incomplete information, players are able to bluff. I am more likely to be successful at poker if I can determine the probability that a competitor is bluffing. This requires a high level of interpersonal intelligence.
When I am seated a poker table in a casino, I look for tells, signs that hint at how a competitor is playing the hand. The more I play in person with other players, the more I will develop interpersonal skills necessary to win. If I play online, however, much of this same information is unavailable to me. I am hampered in my ability to become more emotionally intelligent.
ChaptGPT and similar programs will develop their linguistic intelligence as humans interact with it using natural language. AI programs will not develop the same proficiency in interpersonal intelligence because they are not persons. Humans will not interact with them in the same way as they interact with other humans because we are not wired that way. There will not be a similar feedback mechanism that allows human participants in conversations to develop interpersonal skills. Interpersonal intelligence must be developed in concert with intrapersonal intelligence.
That AI programs will not demonstrate human levels of interpersonal intelligence should not be cause for exuberant celebration on the part of those of us in management roles. Success in executive roles will require as much of a commitment to developing interpersonal and intrapersonal intelligence as is required for all other skills associated with those roles. Those who fail to demonstrate proficiency in these areas will face competition from AI applications for work in areas in which emotional intelligence is not as essential.
Image by Gerd Altmann from Pixabay