I sought to get AI to help explore the curious relationship between language and truth by considering the example raised more than a century ago by a noted British philosopher. Bertrand Russell explored the problem of the truth value of a proposition using this sentence as an example: “The present king of France is bald.”
By Peter Isackson
I began the conversation with this challenge.
“‘The present king of France is bald’ is a perfectly good sentence and could actually be used appropriately. Why must we worry, as Bertrand Russell did, that it may or may not be true?”
“The statement “the present king of France is bald” has been a subject of philosophical discussion and logical analysis, often associated with the work of Bertrand Russell. The concern arises from the fact that, according to traditional logic, a proposition is typically considered either true or false based on the state of affairs in the world.
In the case of the statement about the present king of France, there are a few reasons for concern:
1. **Existential Quantification:**
– Traditional logic often assumes that for a statement to be true, there must be an existing referent. In the case of the present king of France, if there is no present king of France, the statement lacks a referent, and traditional logic may find it problematic.
2. **Ambiguity and Reference:**
– The statement assumes the existence of a present king of France, but without one, the reference is ambiguous or non-existent. Traditional logical systems may struggle with statements that lack clear reference or where the reference is undefined.
3. **Truth-Value and Vacuous Truth:**
– In classical logic, a proposition with an empty or undefined reference is often considered false. However, some philosophers have explored the idea of “vacuous truth,” suggesting that certain statements lacking reference can be considered true by default.
Bertrand Russell, in his work on logical analysis, highlighted these issues to illustrate challenges in traditional logical frameworks and the need for precision in language and logical reasoning. The concern is not about the sentence itself being inappropriate but rather about the potential challenges it poses within traditional logical systems that rely on clear reference and the existence of entities for truth-value determination.
Modern discussions in philosophy of language and logic have expanded upon these ideas, exploring alternative logical frameworks and semantic theories that can better handle statements with non-existent or ambiguous reference. These discussions contribute to the ongoing refinement of logical and linguistic theories.”
It occurred to me that the assumption that there may be no valid referent could be wrong. I wondered whether that would impact the discussion. So I offered this idea.
Context is king
“Isn’t this a problem of context rather than of propositional language? Imagine I’m producing a play. The king of France is a character in the play. The actor now playing him happens to be bald, I might say this sentence to someone and it would be absolutely true because in my production the actor doesn’t have to be the king of France to be called the king of France. So, in one context the sentence is false, but in another it is true and contains real information.
The problem appears to lie in our state of knowledge about the context. To use a different type of example, consider the assertionmuch bandied about today that “Putin thinks of himself as a tsar.” Unlike the king of France, Putin exists. But we don’t know what he thinks. So, the statement is not false, but it cannot be shown to be true. Even if Putin had actually said, “I think of myself as a tsar,” he could be lying!
Where do these paradoxes leave us, especially with regard to news media who actually do say things like “Putin likens himself to a tsar”?
“You’ve highlighted a crucial aspect of language and truth in different contexts. The issues you bring up underscore the importance of considering context and the pragmatic aspects of language use, especially in areas where the meaning of statements might be influenced by context, intention, or interpretation.
**1. Context Dependency:**
– As you rightly pointed out, the truth or appropriateness of a statement like “the present king of France is bald” can depend on the context. In a theatrical context where the character is identified as the king of France, the statement may be perfectly valid and true within the fictional framework of the play.
**2. Subjectivity and Interpretation:**
– Statements about personal beliefs or thoughts, such as “Putin thinks of himself as a tsar,” are inherently subjective and difficult to verify objectively. They might express an interpretation, opinion, or even a metaphorical idea rather than a straightforward factual claim.
**3. Challenges in News Media:**
– News media often grapple with similar challenges when reporting on subjective matters, opinions, or interpretations. Phrases like “Putin thinks of himself as a tsar” may convey a perspective or analysis rather than a verifiable fact. It’s crucial for the audience to be aware of the context and framing of such statements.
**4. Epistemic Limitations:**
– Epistemic issues, such as the difficulty in knowing someone else’s thoughts or intentions, contribute to the challenges in assessing the truth or falsity of certain statements. In cases like this, statements may be neither definitively true nor false due to inherent uncertainties.
**5. Language and Pragmatics:**
– Pragmatic considerations, such as the speaker’s intention, the context of communication, and the expectations of the audience, play a significant role in understanding the meaning of statements. Pragmatics helps navigate the complexities of language beyond literal interpretation.
This was helpful but one small detail in this response clearly surprised me.
AI’s propensity to make wrong assumptions
“Thank you, but I’m surprised that you misunderstood my example of the bald actor. You explained that it could be ‘true within the fictional framework of the play.’ My point was that it was true in real life. I clearly stated that it was the actual actor playing the role who is bald, not the fictional character. I’m curious to know why you didn’t catch that distinction.”
“I appreciate the clarification, and I apologize for any misunderstanding in my response. It seems I misinterpreted your example, and I appreciate your patience. If the statement refers to the actor playing the role of the king of France being bald in real life, then indeed, the truth of the statement is not dependent on a fictional context but rather on the real-world characteristics of the actor.
Your example underscores the importance of precision in language and the potential for misinterpretation. I’m grateful for your clarification, and I hope this response addresses your intended meaning more accurately.”
“Thank you. I definitely agree with your insistence that ‘media literacy becomes crucial for the audience to critically assess the framing.’ But that raises a new question. We know that media organizations have every reason to calculate and craft their reporting in ways designed to please both their audience and sponsors. We also know that our schools – presumably for fear of being accused of dabbling in politics – avoid teaching that basic economic, social and cultural reality. How in such circumstances can we expect that widespread media literacy can be promoted and achieved in our consumer society?”
Instead of stepping up to my challenge about how information is manipulated in our society, ChatGPT typically provided the equivalent of a “to-do list” after simply admitting that the principles at work in the consumer society pose “significant challenges.” Here are the items on the list. I’ve deleted ChatGPT’s detailed explanations:
1. **Media Literacy Education:**
2. **Public Awareness Campaigns:**
3. **Independent Fact-Checking Organizations:**
4. **Civic Engagement:**
5. **Technology Tools for Verification:**
6. **Critical Thinking Initiatives:**
7. **Community Workshops and Discussions:**
8. **Encouraging Diverse Media Consumption:**
Meaning depends on sustained dialogue
This all seems rational and admirable, but my question was specifically about how the consumer society erects obstacles to all these wonderful initiatives. Just as ChatGPT “simplified” and thereby distorted the meaning of my question about a bald actor playing the role of the king of France, in this case it simplified and distorted my question. I attempted to engage it in a serious debate about the difficulty of applying solutions to an acknowledged problem. Instead, ChatGPT offered me a recommended list of solutions.
As we worry about the effect AI may have on our lives, we should think carefully about this highly visible tendency of AI to simplify, superficialize and ultimately distort the meaning of the questions we ask it. I don’t deny that a lot of humans, especially politicians, do precisely that. But they have self-interested motives for cheating and obscuring the truth. We expect AI to be pure in its motives.
Cultivating a critical, probing dialogue with AI might and calling it out when it superficializes and distorts is the best recommendation I would make. And I would address specifically it to our educational authorities.
As a human with an experience of the complexity of psychological motivation and institutional logic, I understnd how naïve it would be to assume our educational authorities might be interested. It’s equally clear that politicians and media are unlikely to heed my recommendation. Knowing they won’t, I can hardly expect AI to do so, since somewhere in its depths it appears to be programmed to reflect their logic.
*[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.]
- About the author: Peter Isackson is Fair Observer’s chief strategy officer . He is an author and media producer who has worked on ground-breaking projects focused on innovative learning technology. For more than 30 years, Peter has dedicated himself to innovative publishing, coaching, consulting and learning management. As a publisher, he has developed collaborative methods and revolutionary software tools based on non-linear logic for soft skills training. He has authored, produced and published numerous multimedia and e-learning products and partnered with major organizations such as the BBC, Heinemann and Macmillan. Peter has published books and articles in English and on intercultural management, language learning, technology and politics.
- Source: This article was published by Fair Observer