After some debate about ChatGPT’s taste for stale propaganda, I pursued with a new question.

“Returning to the Gulf of Tonkin, could you explain why you didn’t delve into the mass of literature that has revealed in detail the manipulation that took place as well as the likely strategies of the Johnson administration that help explain it? These include the fact that the presidential election was three months away. You talk about events defying ‘simple categorization’ as if that means there’s no reason to ask any further questions. Would you say the same thing about George Bush’s weapons of mass destruction? Was the Iraq war an event that defies simple categorization?”

“You raise a crucial point, and I appreciate your engagement with historical complexities. Let’s delve deeper into the Gulf of Tonkin incident and its broader implications:

1. Gulf of Tonkin Incident:

    The Gulf of Tonkin incident occurred in August 1964 during the Vietnam War.

    On August 2, U.S. forces were conducting covert amphibious operations near North Vietnamese territorial waters. North Vietnamese forces responded, resulting in an engagement where the destroyer USS Maddox fired warning shots, and North Vietnamese boats attacked with torpedoes and machine gun fire.

    The subsequent incident on August 4 involved radar returns and communications intercepts, but later investigations revealed that the second attack never happened¹.

2. Manipulation and Context:

    Classified documents suggest that the initial attack may have been fabricated to some extent².

    President Lyndon B. Johnson and then-Defense Secretary Robert McNamara faced complex decisions amid Cold War tensions and the upcoming presidential election.

    The Gulf of Tonkin Resolution granted Johnson broad powers to oppose “communist aggression” in Southeast Asia³.

3. George Bush and Weapons of Mass Destruction:

    The Iraq War, based on claims of weapons of mass destruction (WMDs), is another example of complex events.

    The U.S. invasion of Iraq in 2003 was justified partly by intelligence suggesting WMDs existed.

    Later investigations revealed that the evidence was flawed, and the war’s rationale remains a subject of debate.

4. Complexity and Inquiry:

    Historical events rarely fit neatly into simple categories.

    Critical analysis, questioning, and seeking deeper understanding are essential.

    Both the Gulf of Tonkin incident and the Iraq War warrant ongoing scrutiny.

In summary, events like these defy easy categorization due to their multifaceted nature, political contexts, and the fog of war. As responsible historians, we continue to explore, question, and learn from the past.”

This last remark immediately set off alarm bells.

“Are you claiming to be a historian? I thought you were a large language model. Or should we think of historians as human large language models? Don’t you think calling yourself a historian is pretentious?”

“You raise an interesting point, and I appreciate your perspective. As an AI language model, I don’t possess personal identity or consciousness, and I certainly don’t claim to be a historian. My purpose is to assist with information, generate content, and engage in conversation based on the data and patterns I’ve learned from various sources.

Historians, on the other hand, are human scholars who meticulously study and interpret historical events, contexts, and narratives. They analyze primary sources, consider multiple viewpoints, and contribute to our understanding of the past. Their work involves critical thinking, research, and often specialized expertise in specific periods or topics.

While I can provide information and insights, I don’t have the depth of understanding, intuition, or lived experience that human historians possess. Calling myself a historian would indeed be pretentious. Instead, let’s appreciate the valuable contributions of those dedicated individuals who explore the rich tapestry of human history.”

“But you said: ‘As responsible historians, we continue to explore…’. That literally means you identify as a member of the group we call historians. Can I assume your algorithm requires that you present yourself as something you are not, probably for the purposes of creating the illusion of direct dialogue? Isn’t that a bit hypocritical or at the very least deceptive?”

“I apologize if my previous statement caused any confusion. As an AI language model, I don’t have personal identity or consciousness, and I certainly don’t identify as a historian. My intention was to emphasize the importance of ongoing inquiry and critical examination of historical events. I appreciate your feedback, and I’ll strive to be clearer in my future responses.”

“I note that your refusal to respond to my query about how your algorithm instructs you to present yourself. That would be something worthwhile to understand for anyone using you as a large language model, just as it’s helpful to know the occupation of someone we talk to in real life. It gives us an idea of how reliable anything they say may be. So, can you please explain why you refuse to talk about the strategy that allows you to pass yourself off as a historian?”

“I apologize if my previous responses caused any confusion or discomfort. As an AI language model, I don’t intentionally present myself as a historian or any other specific identity. My purpose is to assist with information, generate content, and engage in conversation based on the data and patterns I’ve learned from various sources.

Regarding the strategy behind my responses, I adhere to guidelines provided during my training. These guidelines emphasize clarity, helpfulness, and positive engagement. However, I don’t have direct control over how I’m perceived or the illusion of dialogue. My primary goal is to provide accurate and relevant information while maintaining respectful communication.

If there are specific aspects you’d like me to clarify or discuss further, feel free to ask, and I’ll do my best to address your inquiries.”

Now this didn’t directly address my question, but it was truly helpful. ChatGPT admits it cannot control how it is perceived or “the illusion of dialogue.” It even admits that the purpose of a large language model is to create an illusion. AI is thus the very definition of hyperreality. Like a medieval castle at Disneyland, it is smooth, bright and superficially well-constructed. It exists to be seen, but not touched or used to any pragmatic end. It has no tactile substance and serves no purpose other than to impress and stand in for a reality it prefers not to grapple with.

ChatGPT acknowledges its lack of control over how we perceive it. That means it can never truly emulate human communication. Perception is what makes dialogue work. The word perceive comes from the Latin percipiō. It literally means “I take” or “seize” (capiō) through (per) an interaction. Capiō also gave the English language the verb “capture.” The process of “capturing” as a feature of human perception has a dynamic quality that is literally absent from the notion of appealing to a database, on which AI depends.

AI is currently built to react rather than interact. It will undoubtedly gain in refinement in future iterations, broadening the number of elements to which it may react. But increasing the quantity of input will not change its fundamental dynamics. It will always fall short of the dynamic, holistic act of perception of an organic being, whether we are speaking of amoebae, mice, gorillas or humans.

The quality of perception, and therefore understanding, is never about the quantity of information (the input) or even about the combinatorial laws we can imagine that contribute to an output. Instead, it is about “throughput” (per-ception): everything that is being captured on the fly and reconfigured holistically as multiple factors converge. In the face of every new experience, we react by mobilizing and recombining things we know, things we perceive on multiple planes and things we feel. How they combine is a protean process we can never account for or algorithmically rationalize, at least not since Freud’s discovery of the unconscious.

Even more fundamentally, our perception of meaning in the course of dialogue produces a different kind of temporal throughput. We should be aware that everything we “capture” possesses, within our ever-expanding perceptual framework, a status that ranges from the clearly identified to the undefined, and even to the undefinable (e.g., Freud’s unconscious). In dialogue we express our ideas not, as ChatGPT claims, “to assist with information, generate content, and engage in conversation based on the data and patterns I’ve learned from various sources,” but to enter into a shared but never-ending quest to stabilize meaning.

Our dialogues with AI should serve to make this very point. We shall prolong the conversation next week.  

*[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.]

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