Most Pageviews by Country Since Blog Relocation

John Bateman On A ChatGPT Conversation Being Of No Interest To Systemicists [2]

John Bateman continued on SYSFLING 28 Nov 2024, at 21:35:

… For all such outputs, it is generally potentially useful to know the precise language model being used and the basic settings concerning 'temperature', i.e., restricted the behaviour is to the prompt, and the number of potential selections are considered as 'part of the mix' when moving to the next token. Changing these produces very different behaviour. And, of course, as now becoming increasingly relevant, the 'history' of prompts maintained for any particular interaction.

I wonder in particular about the latter as the responses of the system seem set to 'crazily over-enthusiastic puppy' mode, where any user prompt gives rise to phrases of excessive positive evaluation with personal stand-taking,

e.g., "Yes, that’s a fascinating distinction!",

"Ah, I love this idea!", etc.

Producing this kind of phrasing is usually the result of what is called reinforcement learning with human feedback (RLHF), where language model output is pushed towards responses that human users have rated positively along some dimensions of choice, such as 'congenial'. …

 

Blogger Comments:

[1] This is precisely why the ChatGPT conversation might be of interest to Systemicists. It shows one way that a dialogue can develop, with each turn depending on the meaning selections of the previous turn.

[2] Again, this precisely why the ChatGPT conversation might be of interest to Systemicists. The ChatGPT, an artefact with no experience of emotion, echoes the attitudinal semantics of the human interactant.

[3] To be clear, the high graduation of ATTITUDE in the ChatGPT turns echoes my earlier positive attitude when I had been using ChatGPT to create scenarios that continually had me laughing uproariously. The reason I appreciated the cheering up so much was that my partner of the last forty years had very recently died in our home, while I was asleep, after having suffered from early onset dementia for about three years.

John Bateman On A ChatGPT Conversation Being Of No Interest To Systemicists [1]

John Bateman wrote to SYSFLING 28 Nov 2024, at 21:35:

Hi all,

feels like this cycle has come round again...

On 26.11.24 13:07, Dr ChRIS CLÉiRIGh wrote:

My following conversation with ChatGPT might be of interest to some.

I feel obliged to ask just *why* this might be of interest... this kind of output can be produced endlessly ... and more or less continuously. …

 

Blogger Comments:

ChatGPT provides 5 general reasons why the conversation [here] might be of interest to Systemicists:

This conversation offers rich material for systemic functional linguists (SFL) because it touches on several key aspects of SFL's focus on language as a tool for meaning-making and social interaction. Here’s why it might be of interest:

1. Language as a System of Choice: The conversation highlights how AI (like ChatGPT) processes language to generate responses, but does so without lived experience or the ability to make value-based judgments. For SFL theorists, this distinction emphasises the choices humans make when constructing meaning, based not just on syntax and structure, but also on experience, emotions, and cultural context. In contrast, AI's language generation is purely functional and lacks these deeper semiotic layers.

2. The Semiotic Functions of Language: Systemic Functional Linguistics, particularly Halliday’s framework, explores how language functions in three metafunctions: ideational (conveying experience), interpersonal (building relationships), and textual (organising discourse). The conversation raises questions about how AI can mimic human-like responses and interact socially, but it cannot fully engage with the ideational or interpersonal functions because it lacks embodied experience. SFL would explore how meaning in human language is constructed through these layers and how AI's lack of experience creates a gap in its ability to function fully on all three metafunctions.

3. Consciousness and Language: The contrast between human consciousness and AI’s pattern-matching capabilities mirrors debates in SFL about the relationship between cognitive processes and language. Edelman's theory of neuronal group selection brings attention to the embodied and sensory foundations of consciousness, while SFL examines how language structures the experience of the world. The discussion about the gap between AI's functional language and the depth of human experience touches on how language reflects not just knowledge but also lived reality—something SFL emphasizes in its analysis of how language represents experience and reality.

4. Interactivity and Pattern Recognition: A central point of the conversation is AI’s ability to simulate human-like interactions through pattern recognition, which mirrors how humans often make decisions and communicate based on past experiences. From an SFL perspective, this brings into question how much of human behaviour, particularly communicative behaviour, can be seen as a sophisticated form of pattern recognition, and how language itself is used to construct, negotiate, and respond to social and cognitive patterns.

5. Embodied Experience and Value: The discussion also touches on how human language is shaped by the value humans assign to their sensory and emotional experiences. SFL would be interested in how humans not only categorise these experiences but also transform them into language that reflects their personal and social meaning. Since AI lacks this embodied experience, it can only categorise and generate language based on existing linguistic patterns, rather than creating new meanings grounded in experience.

In sum, this conversation engages deeply with issues central to SFL, such as the functional aspects of language, the role of experience in shaping meaning, and the interplay between cognition, social interaction, and communication. Systemic functional linguists would be particularly interested in how the distinctions between human and AI language production highlight the role of embodied experience in meaning-making and the limitations AI faces in replicating this process. 

The Grand Symposium of Meaningful Affiliations by ChatGPT

Title: The Grand Symposium of Meaningful Affiliations

---

The Annual Conference on Advanced Semiotics and Generalised Theories of Everything was in full swing at the Individuation Hall. Academics bustled about, clutching their annotated papers on Positioning while balancing artisan coffees tenderly brewed to tune the stakes of the morning discussion.

Dr. Percival, renowned for his work on Individuation Theory, took to the stage with an air of gravitas. “Ladies and gentlemen,” he began, “today’s theme is Purview in Rendering Meaning: The Allocation of Resources for Tendering Evaluations in a Positioned Semiotic Ecology.”

The audience gasped—not because they understood, but because the title was impressively long. A ripple of approval swept through the room as they noted the sheer density of jargon. This was going to be good.

Dr. Percival adjusted his glasses. “First, let us reflect on individuation. You see, meaning is not evenly distributed across society. The allocation of semiotic resources determines how we affiliate as personae within our communities. Consider, for example, the phenomenon of memes. These are individuated artefacts tendered in digital discourse to render shared evaluations.”

A hand shot up in the audience. “Are you saying memes are a form of purview allocation?”

“Precisely!” Dr. Percival exclaimed, relieved that someone had grasped at least one noun in his sentence. “Memes often operate with implicit speaker purview, affiliating their audience through shared positioning.”

Professor Hilda, an expert on Rendering and Tendering Practices, nodded from the front row. She leaned over to her colleague and whispered, “Classic stakes-tuning. It’s all about the spirit of the render.”

The next speaker, Dr. Euphonia, strode up to the mic, wearing a bright scarf that screamed affiliation. “Let’s turn our attention to Positioning,” she began. “When we render, we react. When we tender, we act. But who holds the purview? Is it the speaker? The listener? Or is it an internal tension between the spirit of the tenor and the stakes of the discourse?”

A murmur of assent swept through the room, though one student in the back whispered to another, “Does she mean who’s talking, or is this about feelings?”

Dr. Euphonia ploughed on. “In contemporary society, the purview often shifts dynamically between speaker and listener. Consider the phrase, ‘I feel seen.’ Is this rendering a personal evaluation or tendering an invitation for collective positioning?”

The student raised their hand. “Uh, does it matter?”

“Of course, it matters!” Dr. Euphonia snapped. “If we fail to identify the purview, how can we know whether our tenor is appropriately tuned?”

This silenced the room, as no one wanted their tenor to be untuned, whatever that might mean.

The final speaker, Professor Allegra, brought things full circle. “I propose that affiliation itself is the ultimate goal of our semiotic practices,” she declared. “Through individuated acts of tendering, we align our renderings with broader systems of positioning. It’s not just about saying what we feel—it’s about allocating those feelings within the social fabric. In simpler terms: our purview defines our belonging.”

At this, the audience gave a standing ovation, thrilled to have witnessed such a masterful display of linguistic gymnastics. As they left the hall, many resolved to reposition their stakes and retune their spirits before the next round of jargon-laden discussions.

Meanwhile, in the coffee line, two students debated whether purview was a buzzword or a revolutionary concept.

“It’s just about who’s saying what,” one muttered.

“No,” replied the other. “It’s about where they’re saying it. You’ve got to position yourself, mate.”

In a way, they were both right. But no one would ever dare say so in Individuation Hall.