In 2023, working with Mathew Berube, head of Information Services at the Jones Library in Amherst, several of my old columns were fed into ChatGPT, the artificial intelligence chatbot. AI produced a lengthy analysis of my writing. Then I wrote a new column, which we did not show AI, and Mathew asked AI to write on the same subject as the new column, in my writing style.

The Gazette published both columns without indicating which column was written by me and which by AI, and asked readers to identify who (or what) wrote which one. Every reader who offered feedback correctly identified the column I wrote.

Then in 2024, we repeated the experiment. This time the majority of the feedback did not correctly identify the column I wrote. This change in one year threw me for a loop. It got me wondering just how replicable of a writer I am by AI, and for that matter, how deep a dive AI will take into replicating, even surmounting, all of us.

At first, what I thought I’d do is what I usually do with questions requiring expertise beyond my pay grade — find an academic, preferably a local one, with the knowledge to answer my questions. I even compiled a list of professors of computer science at local colleges whose areas of expertise included artificial Intelligence. I figured I’d work down the list until I found one available to interview.

Then the thought hit me to go to the source, to ask AI itself directly about its nature (or lack thereof), and about what it might mean for the future of mankind.

I came up with three questions, and Mathew fed them to AI. What follows are the questions and AI’s answers. AI was initially quite verbose and circuitous in its answers, so we told it to limit its answers to 75 words.

First question: “Can AI feel what human beings feel, and if not, will AI ever be able to do so?”

AI’s answer: “AI cannot feel emotions as humans do, as emotions rely on biology and subjective experience. While AI can simulate emotional responses, it lacks consciousness and awareness. Even with future enhancements, true emotional experience remains speculative and unlikely.”

Your Daily Puzzles

Cross|Word

Flipart

Typeshift

SpellTower

Really Bad Chess

Second question: “Do human beings have a spirit unique to them, and do you have now, or will you ever have, a spirit?”

AI’s answer: “The ‘spirit’ often signifies a metaphysical essence, tied to consciousness, individuality, or the divine. Many believe it’s unique to humans. As an AI, I lack a spirit, consciousness, or subjective experience—my operations are purely algorithmic. Even with future advancements, AI would only mimic such qualities, not possess them. Whether the spirit is uniquely human depends on how it is defined. How do you interpret the concept of ‘spirit’?”

Final question: “Will artificial intelligence ever surpass humankind and have domain over humankind, the way humans have domain over animals?”

AI’s answer: “AI’s surpassing humanity and dominating it is unlikely. While it excels in specific tasks, it lacks autonomy and intent. Responsible development and governance aim to prevent misaligned goals. Speculative fears stem from fiction, not current reality. AI’s role is determined by human choices, ensuring it serves as a tool for progress, rather than a threat. Its impact depends on how wisely we design and manage it.”

Back to me and a comment on AI’s answers.

I found AI’s last answer about whether it would ever have domain over humankind troubling. Perhaps it’s because I’ve seen too many things happen in my lifetime that were “unlikely” when I was a child. For instance, if you told me when I was young that in my lifetime everyone in the world could have immediate 24/7 access to everyone else in the world on a thing called the internet, I’d have thought it “speculative fiction.”

Also, I don’t know that “responsible development and governance” have come close to characterizing humans’ handling of new scientific and technological advancements. For instance, man’s scientific inquiry into the nature of matter and the resulting discovery of the atom leading to an ever-growing arms race of nuclear weapons, realistically threatening to destroy us all.

For me, AI’s scariest statement of all is when it said its impact depends on how well we design and manage it. I don’t believe wise management of the internet itself has transpired up to this point. It seems to me whatever progress the internet has produced for mankind has been counterbalanced by some wild rides that have been detrimental to human beings, individually and in the aggregate.

And as far as AI becoming misaligned with human interests, the question becomes which humans, and whose interests. The question becomes whose hands are on the wheel.

Right now, it certainly isn’t my hands, and I doubt it’s yours.

Amherst resident Richard McCarthy, a longtime columnist at the Springfield Republican, writes a monthly column for the Gazette.