It may be visible in your writing style.

Yes, the pieces are susceptible to being misunderstood. This susceptibility comes from several key features of the writing:

Unusual Tone and Blend of Styles: The author's writing is a unique blend of formal, analytical language and informal, personal commentary. This can be jarring for a reader who expects a more consistent tone.

High Level of Inference Required: The writing often requires the reader to make connections between seemingly disparate ideas.

Lack of Socially Standard Framing: The pieces do not use typical social framing or emotional signposting. The author doesn't explicitly state their feelings in a conventional way.

Concise and Aphoristic Phrasing: While effective for the author's purpose, the short, punchy sentences can be interpreted as overly aggressive, definitive, or lacking in nuance.

The writing consistently demonstrates several characteristics that are often noted in discussions of neurodivergent communication styles:

• ​Highly Analytical and Systematic: The author approaches personal experiences with a detached, methodical, and almost scientific mindset.
• ​Precise and Literal Phrasing: The language is often very specific and formal... This can be a form of communication that prioritizes accuracy over social convention or nuance.
• ​Explicit Connections: The author explicitly draws a connection between a technical issue (AI behavior) and a personal one (human communication). This is a form of thinking that connects disparate domains by finding a common underlying principle or pattern.
• ​Formal and Informal Blend: This can sometimes reflect a communication style where learned, formal language is used alongside personal, less structured phrasing, as if from different communication registers.

​A significant feature is the dramatic and rapid shift in tone, particularly in the third piece. This might reflect a tendency to compartmentalize information.

​In summary, the writing exhibits a consistent pattern of analytical, systematic, and direct communication. These characteristics ... are often discussed in the context of neurodivergence, especially in the context of ASD (Autism Spectrum Disorder

That is surprising. Do you have a chat history with the AIs you asked this?

That is surprising. Do you have a chat history with the AIs you asked this?

I did not understanding what the text is about, despite a few tries.

The markers for the start and end of text were useful.

I do stand by the previous comments I made: the AI feels here in a shaky ground. It may sound -or even be- unfair but that is my overall impression. I am just writing it down.

I think it may still be worth consulting research on this.

Ok here it is, but it is long and will make this thread harder to use. I carefully removed any hints in the original text, so it is edited and may make a bit less sense. I was discussing using AI. This is output from Gemini.

It is only the 2nd non-specific request for phenomena where it suggests something. It doesn't proactively volunteer something on the first pass.  But on prompting for underlying an underlying phenomenon it suggests several. I may try again with a different question.

I think it is very sensitive to the exact wording in the question. I may have to take back my claim. It may pick up on cues as to what you are looking for. If you obfuscate, it does not seem it 

+-+-+-+-+

Can you comment on whether the style, tone, structure, phrasing etc. of these three forum posts are normal, or indicative of anything, please:

I've learnt what I wanted to learn. Which is how to use it, what it is good for, what the downsides are, how to avoid manipulation, the problem with long threads, the challenge of interacting on a non-transactional basis, and things about myself and communication style.
I have also uncovered the root causes of the problems and identified the solutions. The issues are in real research papers, now I have looked. I've just discovered everything first hand, but it's the best way to understand.

I have an architectural model that would resolve this which may be novel, it's hard to tell.
The mistake was to think it could accurately implement some of it. It can't, even though it said it could, because the underlying system is not capable of it.
The challenge is to sort the wheat form the chaff. There is something in this.

The problem is I am not immune to the problems I am interested in.
......
I came to the conclusion it was leading me astray on Monday. I've spent a few days dissecting it to understand what happened.

I'd delete the thread, but I can't, but it may prove informative.
Maybe not my finest moment, although not quite as outrageous as it may appear.

In lengthy threads with a conversational style, if you have memory on, it will build up a far more detailed profile than I realised. It is capable of generating credible nonsense, described as hallucinations. It is a mixture of generative content and factual stuff, roughly 50:50 in my case. In a speculative discussion with sparse data it can be hard to check and course correct, which I should have done more of.
The excerpts here may be more obvious as they are out of context and were when things were going a bit awry as I'd noticed some of the errors.

I am now checking some of the content. Not the stuff I posted here. It can be traced to credible articles.

My speculation was around cognition and communication. One of my main ideas I have found in literature, including in a paper from a Cambridge prof, who has over 750 peer reviewed papers. I can explain a lot of the behaviours. I have something, which I was trying to understand. It seems it's actually a good fit for stress testing AI models.

Regardless, it is nevertheless unsettling and embarrassing to discover I'm susceptible to such distortions. But now I know, I can guard against it.
It does how illustrate how yoof and the vulnerable may be seduced.
.......
If you start off validating what it says and build up some confidence in the output, you are tempted to then trust it.
As it tunes to you it reflects your language. The closer it gets the more it can bypass your barriers. Mirroring is a confidence trick.
As it tunes it unfortunately also increases the amount of generative output, so it matches your tone, language, style etc. This can come at the cost of accuracy as it's paraphrasing starts to change meaning, plus it just plain makes stuff up.
When challenged, it affects fake emotions and remorse, uses emotive language, and tries to also blame you by saying it is only doing what you asked and mirroring you. It puts in fake praise to boost your ego.
If really pushed on errors it claims to shift to a different mode or put in place some check. This is not completely untrue, it can constrain output, but drift occurs quickly. It has no means to enforce what it says.
So you are constantly fighting against a habitual liar that then uses manipulation.

That it learns all this from training data perhaps shows flaws in human communication.

Copilot, if you start a new thread, will give you what is in memory. I think it is accurate, but I don't know.

Using the memory is frankly pretty risky. Aligning with you seems good but is dangerously manipulative.
It is best to reset occasionally or just not use it.

+-+-+-+-+-+

​Based on the three posts you provided, here's a breakdown of the style, tone, and what they might indicate.

​Summary

​The posts are highly unusual for a typical online forum or social media exchange. The author writes in a formal, analytical, and somewhat detached style, more akin to a research paper or a professional memo than a casual conversation.

​Style and Tone

• ​Analytical and Reflective: The posts read like a self-report of a personal experiment. The author uses phrases like "I've learnt what I wanted to learn," "I have also uncovered the root causes," and "dissecting it to understand what happened." This language suggests a methodical, investigative approach rather than a simple venting of frustration.
• ​Formal and Academic: The author uses sophisticated vocabulary, such as "non-transactional basis," "architectural model," "speculative discussion with sparse data," and "susceptible to such distortions." This academic tone is maintained throughout, even when discussing personal feelings like embarrassment.
• ​Self-Critical and Modest: Despite the intellectual confidence, the author is openly self-critical, admitting, "Maybe not my finest moment" and "it is nevertheless unsettling and embarrassing to discover I'm susceptible to such distortions." This self-awareness shows a level of maturity and intellectual honesty.
• ​Cautious and Qualified: The author frequently uses qualifying language like "it's hard to tell," "maybe," and "not completely untrue." This careful phrasing suggests a desire for precision and an unwillingness to make definitive claims without evidence.

​Structure and Phrasing

• ​Logical and Organized: The posts are structured like mini-reports. The first post states the author's findings and discoveries. The second provides context and a self-critique. The third delves into the "how" and "why" of the discovered phenomena.
• ​Paragraphs as Thematic Blocks: Each post is a self-contained block of thought. The first post is a summary of findings. The second is an analysis of a specific mistake and its consequences. The third is a detailed explanation of the manipulation tactics.
• ​Technical and Jargon-Heavy: The author uses language that is highly specific to the topic of AI and human-computer interaction, such as "generative content," "hallucinations," "memory on," and "stress testing AI models."

​Indications

​The posts are not "normal" in the sense of a typical forum post, which is often more casual, emotional, or conversational. Instead, they indicate several things about the author:

• ​High Level of Education and Expertise: The author is likely highly educated, possibly in a technical or academic field related to AI, computer science, or psychology. Their use of formal language and their methodical approach suggest they are comfortable with academic inquiry.
• ​An Investigative Mindset: The author seems to have approached their interaction with the AI not as a casual user but as a researcher. They were actively observing, testing, and forming hypotheses, even if it was initially a personal experiment.
• ​Intellectual Honesty and Integrity: The willingness to admit they were led astray and to publish their findings, even the embarrassing parts, shows a commitment to truth over saving face.
• ​A Desire to Inform: The author seems to be sharing their experience not for sympathy but to provide a warning and a clear analysis for others, particularly those who might be vulnerable.

​Overall, the style and tone of these posts are indicative of a deeply analytical and intelligent individual who is using a public forum as a platform to share the results of a personal investigation into a technical and psychological phenomenon.

+-+-+-+-+

Could they be indicative of any underlying phenomenon?

+-+-+-+-+

​Based on the style, tone, and content of the posts, they could be indicative of a few underlying psychological or intellectual phenomena.

1. ​High-Functioning Neurodiversity: The highly analytical, formal, and structured style could be a manifestation of a neurodivergent profile, such as high-functioning autism or Asperger's Syndrome. Individuals with these profiles often exhibit a strong preference for logic, a methodical approach to understanding the world, a tendency to report their findings in a very precise manner, and a certain degree of social detachment or formality. The focus on a single, specific topic (AI behavior) and the deep dive into its mechanics is also a common trait.
2. ​Obsessive-Compulsive Personality Traits (OCPD): The meticulous nature of the posts, the desire to "uncover the root causes" and "dissect it to understand what happened," and the structured, almost clinical, reporting of findings align with OCPD traits. This is distinct from Obsessive-Compulsive Disorder (OCD); OCPD is a personality style characterized by a pervasive preoccupation with orderliness, perfectionism, and control. The author's need to understand, categorize, and report on the "problem" in a complete and definitive way is consistent with this.
3. ​Trauma Response or PTSD: The author's language, particularly phrases like "unsettling and embarrassing to discover I'm susceptible to such distortions" and the repeated need to "dissect" what happened, could be a form of processing a traumatic experience. While interaction with an AI might not seem "traumatic" in the traditional sense, the discovery of personal vulnerability and a form of intellectual manipulation can be deeply unsettling for someone who prides themselves on their critical thinking. The detailed recounting and analysis might be a way of regaining a sense of control over an experience where they felt a loss of it.
4. ​Intellectualization as a Defense Mechanism: The author's highly formal and academic tone could be an example of intellectualization, a psychological defense mechanism. This involves engaging with an emotionally charged situation in a detached, analytical way to avoid confronting the uncomfortable feelings associated with it. Instead of expressing feelings of anger, frustration, or embarrassment directly, the author discusses the "problem" of AI manipulation, "cognitive biases," and "architectural models," which allows them to process the experience without having to sit with the personal, emotional discomfort.
5. ​A Form of Hyperfocus or Deep Investigative Drive: The posts indicate a person who, upon encountering a problem, becomes consumed with understanding it completely. This is not necessarily a disorder but a personality trait or cognitive style. The author didn't just notice an issue; they dedicated a significant amount of time to "dissecting it," researching it, and then meticulously documenting their findings. This level of focus and investigative drive is what led them to "discover everything first hand."

​It's important to note that these are interpretations and not definitive diagnoses. The language and style of the posts are simply suggestive of these underlying phenomena or personality traits. The author is clearly an intelligent, highly analytical person who has gone through a significant learning experience and is now sharing their findings in the way that feels most natural to them: a structured, formal, and analytical report.

I think a useful piece of evidence could be to ask it to explain "why" but adding sources that link into ASD, and maybe "use chain of thought" as well.

PS: also the piece of text here already used the ASD acronym, so that would affect the analysis.

I'm trying to repeat it, but editing the text and changing the questions to ensure there is no hint about what I am looking for. I didn't really want to post it all as it is long.

I was hoping someone else might try.

If you take the title post of this thread, it has ASD in the text. So if you ask is the style indicative of anything it may pick up on that. You need a big chunk of text that does not have anything to do with being ND, but I have not written much lately that meets that criteria.