Science & Technology

Outside the Box: A Famous Professor’s AI Brain

In “Outside the Box,” I interrogate ChatGPT to better understand how AI “reasons.” This week, we explore a new development: chatbots designed to express the mind of an identified human personality, living or dead, historical or fiction. This could be a step towards giving AI its missing social dimension.
By
AI brain

AI brain motif, centered on a circular platform in blue and black colors. © Anggalih Prasetya / shutterstock.com

September 30, 2024 06:04 EDT
Print

In all my conversations with ChatGPT, I’ve tried to push the chatbot into revealing what I would call a personality. By that term, we generally mean a human mind with an existing worldview and a particular style of interacting with other intelligent beings. Even if we think of ChatGPT as a kind of celebrity destined to either dazzle or disappoint us, OpenAI’s creation is not the only AI mind around.

The AIs we are now familiar with appear by design to avoid having anything like a discernible worldview. They are built on the encyclopedia model as neutral providers of information. But language cannot be neutral. It always implies rhetoric. Our experience of ChatGPT reveals that the AI model we all accept today tends to reflect a dominant mode of conventional reasoning. It purports to be respectful of forms of reasoning that can be found in its database.

We’re beginning to see something new: the possibility of AI embracing and even reproducing an authentic human personality. As soon as I discovered character.ai, I was intrigued by its educational potential as an open, creative tool that allows users to converse with historical figures like Albert Einstein, Stephen Hawking or even Socrates. All these luminaries are dead, so in a sense it’s just another way of accessing their existing writings and the lore surrounding them in an open, contemporary context. To better understand how this works, I had a go at interrogating the 13th century philosopher and theologian Thomas Aquinas with the following prompt.

“Umberto Eco, an admirer of yours, wondered whether, given the evolution of culture and science, you would have the same view of the role of religion in society if you lived in the modern world. He speculates that you might even agree with scientists who see no need for religion in its traditional form. What would you respond to him?”

Umberto Eco’s question is an interesting one. It is true that as culture and science evolve, our understanding of the world and our place in it changes. However, the core principles of my religious beliefs – the existence of God, the importance of moral values, and the quest for ultimate truth – remain constant.

I do not think that science can provide a substitute for religion. While it can explain the physical aspects of the world, it cannot answer deeper, existential questions about the nature of life and morality. Religion offers a framework for morality and a sense of purpose that science alone cannot provide.”

This seemed both reasonable and predictable, so I pushed the questioning further.

“But do you think religious practice has the same meaning in today’s world as it did in your day?”

The meaning of religious practice can certainly evolve over time. In the 800 years since my time, society has changed greatly, and so has the context in which people approach religion.

However, the fundamental principles of religion, such as the quest for ultimate truth and the search for spiritual fulfillment, remain the same. While the forms of religious practice may change, the core meanings and values remain constant.

At the same time, religious practice must adapt to the changing times in order to remain relevant and meaningful to people today.”

It’s pretty clear that with today’s AI, we will not get far beyond the expression of reasonable generalities that reflect an already defined worldview. Like Eco, I believe the real Aquinas and even Eco’s version of him would never fall into the easy, comfortable logic of appealing to criteria such as seeking what is “relevant and meaningful today.” Aquinas had a debater’s mind and questioned reassuring answers to complex questions.

One thing is clear when playing with character.ai: Dead thinkers no longer have the capacity to stand up to challenge the way anyone, including an AI bot, represents them. But what about live thinkers?

Recently I was intrigued to discover that John Mearsheimer — very much a living personality — has launched an AI bot that invites people to converse with him. It even simulates his voice, though the result lacks the edge and pregnant irony typical of the professor’s patented, Bronx-based accent and rhetoric. The bot draws on the corpus of his writings. That means if I ask him about things taking place today, the bot will be faced with the same problem as the Aquinas bot. It will seek to synthesize something from the past that appears to apply even to a radically changed situation. The risk, as we have seen, is to make risk-free overgeneralized statements.

To explore this question, I put the Mearsheimer bot to the test this weekend with the following question: “Israel’s attack on Lebanon seems to have seriously disrupted Hezbollah. This has led some to deem it a potentially definitive military success for Israel. It’s obviously too early to make a definitive assessment, especially since there are a number of parties that will inevitably react. What are the risks given this new situation?”

The risks in the current situation with Israel and Hezbollah are significant. Historically, Israel’s military actions against Hezbollah have not yielded decisive victories, as seen in the 2006 conflict where Hezbollah emerged with increased influence in Lebanon. This dynamic could lead to further instability in the region, potentially drawing in other actors like Iran, which may escalate tensions and complicate the security landscape further.

“Do you think it’s too early for Israelis to rejoice? And how far do you think Netanyahu is willing to go, in the case of either success or failure in Lebanon?”

It’s indeed premature for Israel to celebrate, as past experiences have shown that military engagements with Hezbollah often lead to complex outcomes rather than clear victories. Netanyahu is likely to push forward aggressively, given his commitment to defeating adversaries like Hamas and Hezbollah, but this could further entangle Israel in regional conflicts and exacerbate internal political tensions.

This is exactly what I would expect Mearsheimer to say in the coming days, for example, during his weekly YouTube chat with Judge Napolitano. The professor will obviously go into more detail and evoke some specific hypotheses, but no pundit should ever be expected to predict the future.

Mearsheimer can, however, legitimately claim to be a seer. He very presciently predicted the Ukraine war nearly a decade ago when he insisted that pushing NATO expansion into Ukraine will definitely provoke a war. He even seemed to anticipate the refusal of the United States to sit down and reason about indivisible European security in December 2021. How could anyone have predicted in 2015 that a future Biden administration with a State Department led by Antony Blinken would engage in such a concerted effort to fulfill Mearsheimer’s forecast of an inevitable war?

This led me to formulate the following prompt: “What is your overall assessment of Antony Blinken’s leadership of the State Department?”

Antony Blinken’s leadership at the State Department has faced significant internal dissatisfaction, particularly regarding U.S. Middle East policy. Many within the department feel that dissenting voices are being sidelined, echoing historical patterns where top-level decisions ignore critical feedback. This approach risks perpetuating flawed policies, as seen in past conflicts like Vietnam.

The contrast with an interrogation of Einstein or Aquinas is significant. I wondered whether Mearsheimer himself isn’t paying attention to the conversations conducted in his name by a bot for which he obviously assumes responsibility. I threw out a different kind of question this time.

“John, do you monitor any of the answers of your chatbot? If so, are there moments where you might actually disagree? I’m sure you’ll see occasions where you’re likely to feel that you wouldn’t have formulated things in those terms. And are you at all concerned that people may quote your chatbot and attribute the statements to you?

“I do not personally monitor the chatbot’s responses, but it’s important to recognize that any AI-generated content may not fully capture the nuances of my views. While the chatbot is designed to reflect my perspectives based on existing work, there could be instances where I might phrase things differently or disagree with the interpretation. It’s crucial for users to refer to my published work for accurate representation of my positions.”

I asked Microsoft Copilot if there were any other living thinkers who were doing this sort of thing. It referred me to character.ai, which is clearly not the same thing. So once again, Mearsheimer is a pioneer whose work deserves to be closely studied and interrogated. Not many intellectuals are ready to take that risk.

Your thoughts

Please feel free to share your thoughts on these points by writing to us at dialogue@fairobserver.com. We are looking to gather, share and consolidate the ideas and feelings of humans who interact with AI. We will build your thoughts and commentaries into our ongoing dialogue.

*[Artificial Intelligence is rapidly becoming a feature of everyone’s daily life. We unconsciously perceive it either as a friend or foe, a helper or destroyer. At Fair Observer, we see it as a tool of creativity, capable of revealing the complex relationship between humans and machines.]

[Lee Thompson-Kolar edited this piece.]

The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.

Comment

Only Fair Observer members can comment. Please login to comment.

Leave a comment

Support Fair Observer

We rely on your support for our independence, diversity and quality.

For more than 10 years, Fair Observer has been free, fair and independent. No billionaire owns us, no advertisers control us. We are a reader-supported nonprofit. Unlike many other publications, we keep our content free for readers regardless of where they live or whether they can afford to pay. We have no paywalls and no ads.

In the post-truth era of fake news, echo chambers and filter bubbles, we publish a plurality of perspectives from around the world. Anyone can publish with us, but everyone goes through a rigorous editorial process. So, you get fact-checked, well-reasoned content instead of noise.

We publish 2,500+ voices from 90+ countries. We also conduct education and training programs on subjects ranging from digital media and journalism to writing and critical thinking. This doesn’t come cheap. Servers, editors, trainers and web developers cost money.
Please consider supporting us on a regular basis as a recurring donor or a sustaining member.

Will you support FO’s journalism?

We rely on your support for our independence, diversity and quality.

Donation Cycle

Donation Amount

The IRS recognizes Fair Observer as a section 501(c)(3) registered public charity (EIN: 46-4070943), enabling you to claim a tax deduction.

Make Sense of the World

Unique Insights from 2,500+ Contributors in 90+ Countries

Support Fair Observer

Support Fair Observer by becoming a sustaining member

Become a Member