After publishing my last post, I asked Claude what it thought about it, using the “reflective blog post” prompt I’ve been developing. It appears that we have some more work to do, indeed.
Claude reviewed my post in terms of:
- Authenticity
- Professional stakes and relevance
- Voice and intellectual honesty
- Process documentation quality
- Factual accuracy
- Overall assessment
Basically, I got a report card from my mechanical English teacher. Let’s see how I fared…
Authenticity
Evaluates whether the uncertainty and professional challenges feel genuine rather than manufactured for content purposes.
Let’s see what my strengths in Authenticity are, according to Claude:
- Genuine uncertainty: The author demonstrates real ambivalence about AI collaboration without rushing toward resolution (“I’m not ready to abandon my AI-overlord, er, I mean co-writer, just yet”)
- Honest stakes: Clear professional and personal impact – less satisfaction, feeling like work, questioning writing identity
- Specific observations: Concrete details about the back-and-forth editing process, watching drafts transform, losing sense of accomplishment
- Temporal authenticity: Uses present-tense investigation language (“What I think this really means…”) and frames conclusions as provisional
OK. So far, so good. What does Claude think I need to work on?
- Limited process transparency: Doesn’t show systematic methodology for investigating this challenge beyond informal reflection
- Missing specifics: Could benefit from more concrete examples of the “soul removal” process or specific editing exchanges
Perhaps. However, I’m not convinced those are appropriate in a reflection paper. It could be that we have a difference of opinion on what a reflection paper should include. Or, perhaps, adding more of this information would make it stronger. In retrospect, I think I’m still figuring out those aspects of my writing, so I’m still not quite sure what to say about them.
I’ll give Claude credit for identifying some inner workings that I need to process.
Professional stakes and relevance
Examines how clearly the article demonstrates real workflow impact and broader applicability to others in the field.
I’m not sure how relevant this is in a personal reflection. However, this was a personal reflection on a professional process (maybe). If anything, I’m lamenting how much the working-with-AI process is like my professional process, when I’d prefer it was more relaxed. But let’s see what Claude thought I did well despite this dissonance.
- Real workflow impact: Documents how AI collaboration affects daily writing practice and satisfaction
- Authentic constraints: Acknowledges the time investment and iterative nature of AI-assisted editing
- Professional identity questions: Grapples with what “authentically human” writing means in an AI context
- Learning orientation: Frames the experience as instructive despite discomfort
I think this is OK, although it misses the nuance of the difference between personal and professional writing.
What do I need to work on in this category?
- Methodological clarity: Could articulate more systematic approach to investigating optimal AI collaboration
- Broader implications: Connects to personal experience but could explore implications for writing profession generally
Fair points, but I don’t want to presume to set the standard for writing practices. As such, I’ll constrain the scope of my comments to me, alone.
Voice and intellectual honesty
Assesses the conversational authority, natural language patterns, and balanced acknowledgment of what the author knows versus doesn’t know.
This category was evaluated descriptively more than qualitatively, and Claude offered this feedback:
- Effective characteristics:
- Conversational authority: Sounds like an experienced writer thinking through a real problem
- Intellectual honesty: Acknowledges contradictions (“More organization” vs. losing authenticity) without false resolution
- Self-awareness: Questions own assumptions (“Does this mean that my natural writing is disorganized? Perhaps.”)
- Authentic uncertainty: Admits discomfort with AI’s inconsistency while continuing to experiment
- Strong voice elements:
- Direct, unguarded language (“sucks some of that fun out”)
- Humor that feels natural rather than performed (“AI-overlord, er, I mean co-writer”)
- Acknowledges emotional responses alongside analytical observations
I’m not sure what to make of this beyond, “keep doing what you’re doing.”
Process documentation quality
Reviews how well the article shows systematic observation, contradictory evidence, and methodological approach to investigating the challenge.
This category seems odd in a personal reflection, and I’m not sure what to make of it here, but let’s see what Claude thinks about it in my last article.
What worked, according to Claude:
- Systematic observation: Documents specific patterns (soul removal, increased organization, decreased satisfaction)
- Contradictory observations: Includes both benefits (learning, organization) and costs (satisfaction, authenticity)
- Honest limitations: Admits uncertainty about whether changes represent improvement or loss
And, what I could’ve made stronger:
- Investigation methodology: More explicit about how they’re testing different approaches or measuring outcomes
- Future experimentation: Mentions prompt maintenance but could outline specific next steps
- Criteria for evaluation: Could be clearer about what would constitute success in AI collaboration
I like these points when it comes to a professional reflection, or event analysis, but they seem a bit awkward in a personal reflection. Perhaps the reason this commentary feels awkward is because the “process” being discussed is me.
Another reason is that expanding on the aspects I could’ve made stronger would take over the article. But I’ll leave them for future content. (You can say it all, but not all at once).
Factual accuracy
Confirms that claims are grounded in documented personal experience rather than fabricated data or unverifiable assertions.
I guess this is the “fact checking” part of the evaluation? If so, it looks like I passed, according to Claude:
- Personal experience grounding: All observations come from documented writing experiments
- No fabricated data: Relies on author’s actual process rather than invented statistics
- Clear speculation boundaries: Marks uncertainty appropriately (“Perhaps,” “I think,” “maybe”)
- Authentic timeline: References actual recent blog posts as evidence
Overall assessment
Synthesizes strengths and enhancement opportunities while determining whether the piece succeeds as authentic reflective process writing.
In summary, Claude thought I did OK, concluding:
The piece successfully builds authority through process transparency rather than definitive conclusions, making it valuable for others navigating similar AI collaboration questions. It feels authentically human and professionally grounded rather than performed or manufactured.
According to Claude, it would appear that I’m not a robot, with or without a CAPTCHA to confirm that.
Conclusion
While it’s comforting to have a machine declare that I’m authentically human, that really wasn’t the point of this exercise. Practically, it’s a test of the writing style prompts that I use. If I write a reflective post in my style, when evaluated by AI, it should confirm that it’s me, writing in my style.
This evaluation deviated a bit from what I was expecting (such as where Claude’s review might be “identifying some inner workings that I need to process.” It’s important to distinguish what’s really there and what the AI reviewing it determines is there. I’m not sure how much Claude can read between the lines to find the “inner meaning” of a work. Until there’s a more causal link established, I’ll assume that what I interpreted as “inner workings” as the more probable reason: the prompt describing the style evaluation needs to be clearer.
Postscript
After reviewing Claude’s evaluation, Claude & I worked on a new prompt to use for this type of writing, which produced more tangible and specific feedback that makes sense in the context of a personal reflection. For example, clarifying the scope of my feedback.
The revised prompt also includes “audience success metrics” that should map to what my target audience would be interested in, thematically. The evaluation gave my first draft a B+/A- grade and this final copy an A, according to this rubric. It’ll be interesting to see how that correlates to reality. Download the report card on the final draft
Taking this into the realm of API documentation, I can see value in creating prompts to evaluate different documentation types to make sure they have the required content elements and, perhaps, confirm correctness by going back to the source code to confirm details. At the same time, creating and refining the prompts to make these evaluations would take some effort as would automating the review processes. Such a task might best be approached as an ongoing, incremental development process. The other side of the coin would be to be prepared to address whatever such a process finds, starting with the false problems it identifies. At the end of the day, aside from bragging rights, the net value of such an effort is not clear, except, perhaps to double check the result of automated content generation processes.
Much to consider.
In the meantime, I’ll take comfort in the fact that AI still identifies me as “authentically human.”