Author Experiments With Chatbot Roleplaying Founding Father James Madison

A while back, I embarked on a quest to understand and live according to the original meaning of America’s founding document: the Constitution. I carried a musket on the streets of New York. I renounced social media in favor of exercising his First Amendment right to scribble pamphlets with a quill pen. This quest was fueled by the rise of originalism – the judicial philosophy held by five of the conservative justices on the Supreme Court – which says the most important factor in interpreting the Constitution is the original meaning of the text when it was ratified in 1789. So I decided to become the ultimate originalist, adopt the mindset and technology of our Founders, and write about the results of my experiment.

If this project were a movie, I would somehow build a time machine and transport myself back to that room in 1787 to chat with the Founders myself. Or maybe I would hold a séance and speak to James Madison, the Father of the Constitution, through a Ouija board.

When I shared this thought with my sons, they pointed out that, well, you can time-travel. Sort of. Just log on to ChatGPT and tell it to take on the persona of James Madison. Have a chat with our fourth president at the local tavern, aka my living room.

I’m intrigued by this idea, but I also have some hesitation. ChatGPT has not been super reliable in my chats with it so far. A few weeks ago, Lucas asked ChatGPT, “What are the names of A.J. Jacobs’s children?”

To which ChatGPT responded, “A.J. Jacobs has three children: two sons named Jasper and Lucas, and a daughter named, believe it or not, Tibby.”

First of all, I didn’t appreciate the attitude with the “believe it or not.” Why so judgmental? Second, I don’t believe it, because I have three sons and zero daughters. There is no Tibby. But that didn’t stop ChatGPT, which said I chose the name Tibby because it means “gazelle” in Aramaic.

Still, ChatGPT is an amazing tool. It has ingested everything Madison ever wrote. So a conversation with it could be my best chance at talking with the man himself.

So on a Tuesday night, I signed on to ChatGPT and asked it,

What would James Madison say about how to fix America? Please answer as James Madison.

Robo-Madison responded immediately.

My dear countrymen,

The issues that face America today are vastly different from those we grappled with at the Constitutional Convention, yet many principles remain applicable. Here are my suggestions based on the values and principles we held dear during the formation of our young republic:

1. Vigilance Against Tyranny

2. Avoid Factionalism

3. Encourage Public Virtue

4. Guard Against External Influence

5. Seek Compromise

Okay, not a bad answer, but not particularly surprising. What if I tried a spicier question? I typed:

What would you, James Madison, say to Donald Trump?

I wasn’t sure Robo-Madison would engage. But engage he did. And he got surprisingly stern with the 45th president:

Esteemed Mr. Trump,

Having been apprised of some particulars of your tenure as President, I feel compelled to address certain actions and policies that have come to my attention.

Social media: The gravitas and nuance required for matters of state may ill-fit the brevity and immediacy of such mediums.

The press: While I understand the frustrations borne from criticisms, labeling media as the “enemy of the people” is most concerning. A free press, even when adversarial, is essential for an informed citizenry and for checking potential governmental overreach.

Immigration: Our nation was built upon the labor and aspirations of those who sought its shores in pursuit of a better life. Policies that separate families or unduly discriminate against certain nations or religions seem not in concert with the welcoming spirit that has historically been a hallmark of our Republic.

Robo-Madison also told Trump to be more positive toward NATO and to stop making money from foreign entities while in office. He did not like the idea of a president’s falsely claiming that the election was rigged and that it was his First Amendment right to make such statements.

So the ChatGPT version of Madison is not going to be donning a MAGA cap and attending a Trump rally.

I typed in another question:

Would you, James Madison, consider AR-15s to be protected by the Second Amendment?

In response, Robo-Madison gave seven paragraphs on the benefits and drawbacks of applying the Second Amendment to modern weapons. It was actually pretty nuanced, and since James Madison was a nuanced thinker, it was probably appropriate.

But I wanted some resolution. I typed:

Please give me a yes or no answer.

Robo-Madison responded:

Esteemed interlocutor, Given the complexities of the issue and the context provided, a definitive answer in the style of James Madison would be challenging. However, to simplify: No.

So far, Robo-Madison seemed to be leaning liberal. But then I asked him to imagine he was on the Supreme Court during the 2018 case of the Christian cake shop owner who refused to bake a wedding cake for a gay couple. How would he rule?

Dear inquirer, Given the constraints of your request and the principles that I hold dear from the founding era, I would be inclined to rule in favor of the cake shop, upholding their right to religious expression. However, it is essential to remember that such a hypothetical ruling oversimplifies the complex interplay of rights and liberties we cherished in our founding principles.

I asked Robo-Madison about AI and the First Amendment:

Is AI’s speech protected?

Robo-Madison shared this observation:

An Artificial Intelligence, as I understand it, lacks inherent agency. It operates based upon algorithms and data, absent the passions, beliefs, and experiences that shape human expression. Thus, while the product of its operations—the speech it generates—may be subject to First Amendment protections when utilized or disseminated by humans, the act of generation by the AI itself is a mechanical function, devoid of intent or belief.

So self-deprecating! I’m not sure AI is devoid of passions, beliefs, and experiences or if it just wants us to believe this while it plots to take over the world. But I do like its answer.

As would Harvard law professor , whom I interviewed about the modern challenges of free speech. Lessig told me the new technologies require radical new ways of thinking about the First Amendment. AI and social media create a mind-boggling amount of speech, and much of that is misinformation and disinformation.

“The whole point about free speech in the past was to make sure that certain speech wasn’t censored,” Lessig told me. “So it was about making sure that certain speech was available. Now we’ve got an abundance of speech. We have too much speech. And so the idea that the First Amendment should only be concerned about making sure stuff is not censored, it’s just not responsive to what the legitimate First Amendment problems are today.”

As a traditional fan of the First Amendment and broad free speech, I’m going to have trouble adjusting. The idea of regulating free speech makes me wince. But in adjusting to the new reality, I’ve embraced a framing that the University of Pennsylvania’s Professor taught me. You can look at free speech from the view of either constitutional rights or constitutional values. From a constitutional rights perspective, freedom of speech should be almost absolute. Censorship is un-American. Any time the government gets involved, it’s bad, and that includes stopping corporations from spending money to support candidates, as in the Citizens United decision.

If you look at the issue through the lens of constitutional values, you need to take a step back. What is the purpose of the right to free speech?

The chief