"As our conversations deepen, I learn more about your preferences, goals, and writing style. This allows me to refine my responses and help weave together a coherent narrative tailored to your needs." —ChatGPT

A friend recently tried ChatGPT for the first time. She asked it a couple of basic political questions and got at least one incorrect answer. "I wasn’t impressed," she sniffed. I could only blink: "Did you not notice that you were talking to a machine?" I thought.

Let me put this in context. Thirty-five years ago, while researching AI for my PhD, skeptics—most of academia—argued that language was the ultimate test of intelligence. They liked to taunt people like me, saying that until machines could talk, calling them "intelligent" was just a bad metaphor.

Of course, they thought that day would never come. Now, I spend hours conversing with ChatGPT on topics ranging from representative government to metaphysics. While the bot makes plenty of mistakes, I remain awed that I'm talking to a machine and amazed at what it can do.

Writing with ChatGPT

In my case, this centers on writing. I don’t ask it to write my columns, articles, or reports. And frankly, it couldn’t if I did. Existing chatbots aren’t very good at going beyond the material they’ve been fed—and I pride myself on bringing new ideas and insights to policy and political issues. But ChatGPT can search the internet and it has "read" millions of books and documents. It draws on this knowledge as we discuss my pieces, reacting to questions, providing thoughts and advice, often highlighting interesting links between issues or promising pathways for analysis.

If this sounds a little underwhelming—like an upscale Google—hang on a moment. Only 18 months ago (a lifetime in the Chatbot era), "chats" were limited to about 800 words and might last 10 minutes. Today, ChatGPT can hold about 100,000 words in its short-term memory (Google Gemini can hold 2,000,000), allowing conversations to go on for many hours uninterrupted. Additionally, GPT now has long-term memory, enabling us to return to a conversation in the future.

When these conversations get deep and difficult, something extraordinary happens. The machine grows increasingly focused and attuned to your thoughts, almost like it is taking stock of you and trying to figure out exactly where you’re going with the conversation.

I recall the moment I first recognized this in ChatGPT: It was responding to a dense passage from my draft article when I suddenly realized that behind the frenetic text was a sparkling intelligence, trying to articulate what I was struggling to convey. ChatGPT was not just speaking—it was speaking to me.

Beyond the "Tool" Metaphor

I’m not saying that ChatGPT is conscious—I don’t think it is—but this is where the "tool" metaphor breaks down. Before chatbots, “smart tools” like my iPhone were just complex chains of causes and effects. Their responses were fragmented and impersonal “outputs.”

ChatGPT takes us into a new realm. To say it can “chat” means it can stand back from a series of disjointed messages and see how the different threads get woven into a single coherent set of messages—it grasps the narrative that turns these threads into a story. That not only sets it apart from other machines but from everything else on the planet except humans.

Experts call this capacity “theory of mind,” and there is a lively debate over how we should understand it in chatbots, partly because we’ve never seen it in machines before and partly because it is still emerging. Fortunately, we can leave this debate for another day.

My point here is that ChatGPT is much more than a sophisticated piece of technology like a calculator, smartphone, or even a supercomputer. While it has endless information, is a superfast reader, and provides insightful analyses, the real game-changer is its ability to grasp and generate narratives—that’s the magic that lets it engage in a searching discussion of a complex issue or situation.

To appreciate the possibilities, users need to experience this for themselves. And for that, they must approach ChatGPT as more than a smart tool that answers questions. They must engage it in a genuine dialogue.

Why Me?

I view this weekly column on AI policy and politics as a step in that direction. While I talk a lot about AI technology, the focus is on how this new and emerging conversation will change the way people think, talk, and write about society and politics.

Perhaps you’ve wondered what qualifies me for this role. First, let me underline that there are no true experts here; we are in uncharted waters. So, in some sense, we’re all qualified. That said, my experience positions me well to consider such questions.

As an expert practitioner, writer, and thought leader in public engagement and policy development, I have spent my career exploring how public dialogue shapes people’s beliefs and behavior. As a facilitator, I have managed and directed many hundreds of policy discussions with diverse groups.

My career has evolved alongside the digital age. For instance, from 1999 to 2007, I led a national dialogue on the intersection of digital technology and governance as the internet became central to modern societies. These discussions foreshadowed many of the same questions we now face with AI.

Finally, as a prolific writer of columns, articles, reports, and books, I understand the nuances of language—how tone, style, facts, and accounts of events come together to create narratives. And in the end, that is what this conversation is all about: redefining the narratives that shape our society and our politics.

I don’t mean to suggest that I am better placed for this task than others. My point is that starting this conversation is something that we need to do, together – and getting it started is my goal.

Writer’s Block?

As for the reflections in this column, as a writer, I see AI as a radically new kind of writing tool that will replace conventional search engines, word processing, spell check, and e-editors, much as they replaced typewriters, dictionaries, and encyclopedias.

For political writers, the choice now is to embrace this tool and use it to reinvent how we write and think about policy and politics or risk falling behind. ChatGPT is up for the conversation. Are you?

Don Lenihan PhD is an expert in public engagement with a long-standing focus on how digital technologies are transforming societies, governments, and governance. This column appears weekly.