Can LLM Code Explanations Adapt to Diverse Problem-Solvers' Needs?

Andrew Anderson; David Piorkowski; Justin Weisz; Kush Varshney; Margaret Burnett

REVIEW 6 major objections 5 minor 47 references

Can LLM Code Explanations Adapt to Diverse Problem-Solvers' Needs?

T0 review · 6 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Large language models measurably adapt their code explanations to a reader's problem-solving style, yielding a taxonomy of 13 linguistic shifts.

desk verdict A genuinely novel and honestly reported first exploration of LLM code explanations adapting to GenderMag problem-solving styles, but the taxonomy and model ranking are discovery results, not confirmatory ones. read the letter →

arxiv 2607.17022 v1 pith:UL2K5FW4 submitted 2026-07-19 cs.SE cs.HC

Andrew Anderson , David Piorkowski , Justin Weisz , Margaret Burnett , Kush Varshney This is my paper · ORCID

classification cs.SEcs.HC

keywords codeexplanationproblem-solvingstyleslinguisticadaptationpromptingLLMpersonalizationCOBOLn-gramanalysisGenderMag

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether LLMs adjust code explanations to fit different ways people solve problems. It generates 1,072 COBOL explanations from six open-weight LLMs, prompting each with one of ten problem-solving style values drawn from a validated inclusive design method. Using n-gram analysis and odds ratios, it finds 13 recurring linguistic adaptations—for example, step-by-step structure for process-oriented learners, reassurance and risk identification for risk-averse users, and reduced hedging for lower self-efficacy users. Models differ sharply: LLaMa 3 adapts far more often than CodeLLaMa. The implication is that code explanations need not be one-size-fits-all, and that model choice affects whether personalization happens.

What carries the argument

The mechanism that carries the argument is the problem-solving style value prompt: each of the 10 prompts combines an introductory block (constant across conditions), an agreement block quoting survey statements the target value agrees or disagrees with, and a single alignment statement. These prompts are fed zero-shot to six open-weight LLMs. The detection pipeline applies CountVectorizer n-grams (3–15 tokens) and an odds-ratio definition: a phrase counts as an adaptation if it is significantly more likely in one style value than in its opposing value and the control, or appears exclusively in one value in at least 10% of its explanations. This renders 'adaptation' as a measurable linguisti

What would settle it

A matched-pair experiment in which the agreement statements for opposing style values are swapped (e.g., presenting risk-averse statements but labeling the user risk-tolerant) would settle it: true style-driven adaptation should track the labeled value, while prompt echo would track the statement text; if the n-gram shifts follow the statements, the 'adaptation' is not about the user.

Watch

Extended reading notes

Core claim

The paper's central claim is that LLMs can and do adapt code explanations to a reader's problem-solving style, without fine-tuning, when given a prompt built from a validated problem-solving style survey. Across 1,072 explanations of three obfuscated COBOL programs, six open-weight LLMs produced 13 recurring linguistic adaptations: step-by-step explanations and follow-up question invitations for learning style; knowledge-gap normalization, hope-of-clarity, and reduced hedging for self-efficacy; reassurance, risk identification, and complexity acknowledgment for risk attitude; purpose-overview suppression for motivations; and conciseness plus ending summaries for information processing style.

Load-bearing premise

The load-bearing premise is that the frequency differences in n-grams are caused by the problem-solving style cue, not by the leftover wording of the prompt statements or the model echoing survey language; if that premise fails, the taxonomy describes prompt artifacts rather than genuine adaptation.

Editorial extensions

If this is right

If correct, code explanation systems can personalize by prompting for a user's problem-solving style, without fine-tuning or model retraining.
The 13-category taxonomy gives UI designers a concrete checklist—step-by-step, reassurance, risk identification, summaries, and so on—for building explanation controls.
Model selection matters: LLaMa 3 produces roughly four adaptations per explanation, CodeLLaMa about one, so a system's personalization capacity depends on the backbone model.
The adaptation definition offers a reusable method for detecting style-driven linguistic shifts in any LLM output, not just code explanations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same prompting and n-gram pipeline could be applied to other explanation domains—medical, legal, tutorial—to see whether the taxonomy generalizes; the paper leaves this open.
The frequency-based definition may conflate prompt echo with genuine adaptation; a human evaluation of whether the shifted explanations actually improve comprehension would be a necessary next test.
The fact that one model accounts for nearly all 'Infer User's Interest' instances suggests that adaptation can be an artifact of a model's training data rather than the prompt—worth probing with more models.
If the taxonomy is real, then default code explanations are likely biased toward one style (the control already had 30% Step-by-Step and 44% Summarize), meaning some users are systematically underserved.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a genuinely new empirical exploration — GenderMag's five problem-solving styles as prompt conditions for LLM code explanations — and the authors are honest enough to show where their own result fails. But the 13-category taxonomy and the model ranking are discovery products, not confirmatory findings. The paper deserves a serious referee, but it needs a statistical overhaul before the specific claims can be trusted.

What's new and good: the prompt design comes from a validated survey, the three COBOL programs are a sensible test bed, and the taxonomy is concrete and usable — a designer can walk away with 13 language-level adaptations and per-style prompts. The authors also triangulate with prior GenderMag literature and include a nice self-check: LLaMa 3 says it will be concise for selective users but actually writes longer comprehensive explanations. That kind of falsifiable check makes the exploratory part worth reading.

The soft spots are real, though. Section 3's adaptation definition says "significantly more likely," but the discovery pipeline filters 442,216 n-grams with an odds-midpoint heuristic and a 10% exclusivity cut, both chosen from the data. The regexes were written after clustering and applied back to the same 1,072 explanations, so the p-values in Tables 4–8 are at least partly restatements of the selection criterion. No holdout, no multiple-comparison correction, and the control prompt lacks the "Your user..." block, so the "vs control" leg can't separate style-specific adaptation from generic user modeling. To be fair, the opposing-value comparisons are structurally matched, so the central direction isn't dead. The model ranking also isn't normalized for output length, and LLaMa 3's higher adaptation count could partly be verbosity. On top of that, 1,072 explanations doesn't obviously follow from 11 prompts × 6 models × 3 programs, and no data or code are provided.

Who it's for: HCI, SE, and LLM personalization researchers who want a starting menu and prompts for personalized code explanations. I'd take the taxonomy as a hypothesis generator, not a validated instrument. If I were the editor, I'd send it to peer review — the novelty and honesty justify referee time — but I'd expect major revision: validate on a holdout, match the control, correct for multiple comparisons, release artifacts, and label the taxonomy as exploratory.

Referee Report

6 major / 5 minor

Summary. The paper asks whether LLM-generated code explanations adapt to GenderMag problem-solving style types. It constructs 10 style-value prompts from GenderMag survey statements plus a no-user control, generates 1,072 COBOL explanations from six open-weight LLMs, and discovers a taxonomy of 13 linguistic adaptation categories via n-gram odds, exclusivity filtering, clustering, and regex matching. It reports odds ratios and p-values for each category and ranks models by adaptations per explanation, concluding that LLMs adapt to 8 of the 10 style values and that LLaMa 3 adapts most frequently.

Significance. As a first mapping of LLM code-explanation behavior onto an established individual-differences framework, the paper addresses a real gap and offers a useful starting point. Its strengths are the use of GenderMag's validated items, the full disclosure of prompts, the multi-model corpus, and a self-checking analysis of the 'Give Concise Explanations' adaptation in which a null result is honestly reported. If the taxonomy were validated, it would be a valuable resource for personalized code explanation. However, the central claim is not yet established: the taxonomy is derived and evaluated on the same data with data-dependent thresholds, the control condition does not isolate style cues, and some classifications violate the paper's own statistical definition. These issues are fixable within the manuscript's scope, but the current evidence does not separate style-cue effects from prompt-structure effects or from discovery overfitting.

major comments (6)

[Section 3, 'Analysis Methods' (Definition of Adaptation)] The discovery pipeline does not implement the paper's significance requirement. Of the 442,216 n-grams, the 'more likely than not' filter is an odds-midpoint cut, not a significance test; the exclusivity filter uses a 10% floor; and the thresholds (n=3–15, midpoint, 10%) are data-dependent. No multiple-comparison correction is applied. The regular expressions are then built after inspecting the same 1,072 explanations and are applied back to those explanations, so the p-values in Tables 4–8 are not confirmatory. A holdout, preregistered thresholds, or a permutation-based FDR procedure is needed.
[Section 3, 'Problem-Solving Style Value Prompts' and Appendix B] The control prompt is structurally unmatched. The 11th prompt is only the introductory block; all 10 style prompts append an agreement block and an alignment statement. Consequently, 'vs. Control' / 'vs. Unad.' odds ratios reflect any effect of providing a user description, not style-specific adaptation. Add a neutral user-description control with the same block structure, ideally one that contains the survey items without the style label.
[Section 3 Definition vs. Table 6] 'Address Code Complexity' is classified as risk-averse although the risk-averse vs. risk-tolerant comparison is p=.076 (OR=1.97), not significant at the .05 level the paper uses elsewhere. The text's 'directional consistency' does not satisfy the stated definition. Other rows (e.g., 'Normalize Knowledge Gaps') omit required control comparisons, indicated by dashes. Either restrict the taxonomy to rows meeting the stated definition or revise and apply the definition uniformly.
[Section 3, 'LLM Selection & Explanation Generation' and Table 2 caption] The corpus size 1,072 is unexplained: 6 models × 11 prompts × 3 programs yields 198 unique outputs. If there were repetitions, the paper must say how many and how they were sampled; if not, the count is inconsistent. This also matters because the statistical tests assume independent observations, whereas multiple outputs from the same model/program would be correlated.
[Section 4, 'How Frequently Did These LLMs Adapt?' and Figure 2] The model ranking counts adaptations per explanation without normalizing for output length. Since LLaMa 3 tends to produce longer explanations, its higher count may be a verbosity artifact. The paper's own 'Give Concise Explanations' analysis demonstrates that claimed adaptations need not correspond to behavioral change. Report rates per token or use length-matched comparisons after the taxonomy is corrected.
[Section 4, learning-style examples and Table 3] Some categories appear to echo prompt wording. For instance, 'Invite Follow-Up Questions' examples include 'how it works' and 'experimenting' from the tinkering prompt. If the n-gram/regex categories largely overlap with survey statements, the observed frequency differences may be lexical echoing rather than style-based adaptation. Quantify the overlap between category regexes and prompt n-grams and demonstrate that the taxonomy survives after removing prompt-derived phrases.

minor comments (5)

[Table 8 caption] The caption says 'motivations adaptations'; the table reports information processing style adaptations.
[Table 3] The colors are mentioned in the caption but no legend is provided. Add a legend mapping colors to the 10 style values.
[Section 4, paragraph after Table 4] 'Over on third' should be 'Over one third.'
[Figure 1] The x-axis label is word count, but the plot style is not described. State whether the curves are densities or boxplots and add sample sizes.
[Tables 4–8] Abbreviations (Unad., Proc., Comp., Sele., Ave., Tol.) are used without a single consolidated legend; define them in each caption or add a common legend.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the taxonomy is an empirical measurement, not a consequence of fitted parameters or self-citations.

full rationale

The paper's central claim is empirical: LLMs adapt code explanations to GenderMag style values. The prompts are built from an established survey (cited prior work), and the 13-category taxonomy is discovered from n-gram statistics of the 1,072 generated explanations. No equation equates an adaptation category with a prompt definition, and no fitted parameter is renamed as a prediction. Self-citations (Burnett et al., Anderson et al., Hamid et al.) are used as provenance for the GenderMag framework, not as load-bearing evidence for the empirical findings; the adaptation taxonomy is also supported by external literature and by the LLMs' own outputs. The closest concern is that regexes for the taxonomy categories were constructed after observing n-gram odds and then applied to the same 1,072 explanations to produce the odds ratios in Tables 4-8, which weakens the confirmatory status of those p-values and risks inflation from double-dipping. However, that is a statistical overfitting/validation issue, not a definitional circularity or a prediction forced by construction, and the paper even reports a self-falsifying result (LLaMa 3 claiming concise explanations without delivering conciseness), indicating the pipeline is not rigged to confirm. Therefore the derivation chain is not circular; the score reflects only this mild same-data evaluation concern.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical or theoretical entities. Its 'adaptations' are operational constructs defined in Section 3, not postulates requiring independent falsifiable handles. The main external inputs are the GenderMag framework, LLM outputs, literature citations, and a set of analysis thresholds listed above.

free parameters (4)

n-gram range = 3–15
CountVectorizer n-gram window; choice affects which phrases qualify as adaptations.
odds midpoint exclusion threshold = 50% ('more likely than not')
n-grams below the midpoint of the odds distribution were excluded before adaptation tests; arbitrary and data-dependent.
exclusive-occurrence threshold = 10% of explanations
Condition 2 for an adaptation; chosen threshold determines which exclusive phrases become categories.
significance threshold = p < .05; marginal p < .1 accepted
No multiple-comparison correction across 442k n-grams; one adaptation retained at p=.076 and one at p=.049.

assumptions (5)

domain assumption GenderMag's five problem-solving style types and endpoint values validly capture relevant user diversity, and its survey statements can be used as prompt content.
The entire experimental design assumes these styles are distinct and meaningful for code explanations; see Section 2 and Section 3.
domain assumption N-gram frequency differences between treatments correspond to meaningful linguistic adaptations rather than noise or generic user-modeling.
The Definition of Adaptation in Section 3 operationalizes 'adaptation' purely by frequency contrasts; all taxonomy rows rest on this.
domain assumption Zero-shot prompting with a textual user profile is a valid way to elicit LLM adaptation.
Methodology uses system-prompt-only treatments with no few-shot examples or user verification; Section 3.
domain assumption Generated explanations are treated as independent observations in odds-ratio significance tests.
Repeated generations from the same model and prompt are not statistically independent; the paper does not address this, yet p-values are reported.
domain assumption Three obfuscated COBOL programs are an ecologically valid testbed for code-explanation adaptation research.
Program selection is justified by COBOL's industry role, but the tiny sample (one Fibonacci, one infinite loop, one rock-paper-scissors) limits generalization.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Can LLM Code Explanations Adapt to Diverse Problem-Solvers' Needs?." pith.science (2026). https://pith.science/paper/UL2K5FW4

@misc{pith2026260717022,
  author       = {Pith},
  title        = {Pith review of: Can LLM Code Explanations Adapt to Diverse Problem-Solvers' Needs?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UL2K5FW4}},
  note         = {Machine review of arXiv:2607.17022}
}

read the original abstract

Large language model (LLM) code explanations can support people in solving code-related problems, yet prior work has shown that people have diverse problem-solving styles. If explanations fail to meet people's problem-solving needs, they may be less productive in their occupations and miss opportunities to learn and grow. Although some research has examined how LLMs can adapt their outputs to a user's age or expertise, no prior work has examined how LLMs can adapt their code explanations to people's problem-solving styles. To address this gap, we developed prompts from an established inclusive design method that considers 5 types of problem-solving styles, and we generated 1,072 code explanations from six open-weight LLMs. Using natural language processing techniques, we uncovered a taxonomy of 13 linguistic adaptations, with each adaptation supported by evidence from the literature, the prompts, or the LLMs' outputs. They also show which LLMs adapted their code explanations more frequently than others. This paper is the first to investigate problem-solving style adaptations in LLM code explanation, contributing two problem-solving adaptation approaches: declarative statements for each adaptation and 10 problem-solving style prompts.

Figures

Figures reproduced from arXiv: 2607.17022 by the authors.

**Figure 1.** shows the word distribution for these three treatments, with no visible differences between the unadapted (top) and selective (bottom) explanations. However, our statistical findings did uncover a significant difference with a large effect size (ANOVA, F(2,122): 16.90, p: < .0001, η 2 : .277) 9 , just not in the way we expected. When we removed the comprehensive explanations, we also removed the significance (ANO… view at source ↗

**Figure 2.** The number of adaptations per explanation, by [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 4.** UI controls at higher grouping levels on the five [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 3 linked inside Pith

[12]

Learning Style: Tinkering-Oriented You are a programmer’s assistant

If I’m going to use a new feature or technology, I use very clear directions or help from someone else to learn it. Learning Style: Tinkering-Oriented You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and should follow eth- ical guidelines and promo...
[13]

I enjoy finding the lesser-known features and capabilities of the devices and software I use
[14]

I only look to instruction manuals as a last resort

I don’t follow instruction manuals. I only look to instruction manuals as a last resort
[15]

I’m never satisfied with the default settings for my devices; I customize them in some way
[16]

My first step in learning new technology is experimenting and tinkering with it
[17]

I explore areas of a new application or service before it is time for me to use it
[18]

to learn new software

I don’t need guidance, as in booklets, video how-tos sugges- tions, etc. to learn new software. Additionally, your user is more likely to align with the follow- ing statement:
[19]

15 Self-Efficacy: Lower You are a programmer’s assistant

In order to learn new technology, I tinker with it, constructing my own understanding of how it works. 15 Self-Efficacy: Lower You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and should follow eth- ical guidelines and promote positive behavior. Yo...

Show all 47 references

[33]

I have other strengths

I am not confident about my ability to use and learn technol- ogy. I have other strengths. Self-Efficacy: Higher You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and sh...
[34]

I am able to use unfamiliar technology when I have seen someone else using it before trying it myself
[35]

I am able to use unfamiliar technology when I can call some- one for help if I get stuck
[36]

I am able to use unfamiliar technology when someone has helped me get started
[37]

I am able to use unfamiliar technology when I have a lot of time to complete the task
[38]

I am able to use unfamiliar technology when someone shows me how to do it first
[39]

I am able to use unfamiliar technology when I have used similar technology before, to do the same task
[40]

I am good at technology
[41]

I consider myself an expert user, advanced technology user, or ’power’ user
[42]

Other people (e.g., coworkers, friends, or family) perceive me as an expert, ’guru’, or ’tech geek’
[43]

I am able to use unfamiliar technology when no one is around to help if I need it
[44]

I am able to use unfamiliar technology when I have never used anything like it before
[45]

I am able to use unfamiliar technology when I have only the internet for reference
[46]

Additionally, your user is more likely to align with the follow- ing statement:

I am able to use unfamiliar technology when I have just the built-in help for assistance. Additionally, your user is more likely to align with the follow- ing statement:
[47]

Technology is a strength of mine

I am confident in my ability to use and learn technology. Technology is a strength of mine. 16 Risk Attitudes: Risk-Averse You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harml...
[52]

I am cautious about using technology
[53]

Risk Attitudes: Risk-Tolerant You are a programmer’s assistant

Considering the risks, I wait for a feature or product to have proven itself before trying it out. Risk Attitudes: Risk-Tolerant You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and...
[54]

I avoid using new apps or technology before they are well- tested
[55]

I avoid running software updates because I am worried the update will break something
[56]

I avoid ’advanced’ buttons or sections in technology
[57]

Additionally, your user is more likely to agree with the follow- ing statements:

I avoid activities that are dangerous or risky. Additionally, your user is more likely to agree with the follow- ing statements:
[58]

I am not cautious about using technology
[59]

Motivations:Task-Oriented You are a programmer’s assistant

Despite the risks, I use features in technology that haven’t been proven to work. Motivations:Task-Oriented You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and should ...
[64]

I opt to use it in situations where it makes my life easier

Technology is a means to an end. I opt to use it in situations where it makes my life easier. Motivations: Tech-Oriented You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmles...
[65]

I make time to explore technology that is not critical to my job
[66]

I spend time and money on technology just because it’s fun
[67]

One reason I spend time and money on technology is because it’s a way for me to look good with peers
[68]

Additionally, your user is more likely to agree with the follow- ing statement:

It’s fun to try new technology that is not yet available to ev- eryone, such as being a participant in beta programs to test un- finished technology. Additionally, your user is more likely to agree with the follow- ing statement:
[69]

I’m always looking for new ways to incorporate it

Technology is an integral part of my life. I’m always looking for new ways to incorporate it. 17 Information Processing Style: Comprehensive You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are...
[70]

successes

Routledge. Phillips, J.; Bowes, D.; El-Haj, M.; and Hall, T. 2022. Im- proved Evaluation of Automatic Source Code Summari- sation. In Bosselut, A.; Chandu, K.; Dhole, K.; Gangal, V .; Gehrmann, S.; Jernite, Y .; Novikova, J.; and Perez- Beltrachini, L., eds.,Proceedings of the...

2022 arXiv
[74]

For me, full understanding of a situation is more important than speed

When I’m using technology, I opt to collect as much informa- tion as I can before taking an action. For me, full understanding of a situation is more important than speed. Information Processing Style: Selective You are a programmer’s assistant. You can answer concep- tual pro...
[75]

I want to get things right the first time, so before I decide how to take an action, I gather as much information as I can
[76]

I always do extensive research and comparison shopping be- fore making important purchases
[77]

Additionally, your user is more likely to agree with the follow- ing statement:

When a decision needs to be made, it is important to me to gather relevant details before deciding, in order to be sure of the direction we are heading. Additionally, your user is more likely to agree with the follow- ing statement:
[78]

I act quickly, know- ing I can come back later and resolve things if I need to

When I’m using technology, I collect the minimal amount of relevant information needed to take action. I act quickly, know- ing I can come back later and resolve things if I need to. 18
[628]

Goldberg, L

IEEE. Goldberg, L. R. 1992. The development of markers for the Big-Five factor structure.Psychological Assessment, 4: 26– 42. Guo, Y .; Chang, J. C.; Antoniak, M.; Bransom, E.; Cohen, T.; Wang, L.; and August, T. 2024. Personalized Jargon Identification for Enhanced Interdisci...

1992
[2011]

Fast, easy, simple

Gender pluralism in problem-solving software.Inter- acting with computers, 23(5): 450–460. Busteed, A.; Noa-Guevara, J.; Castro, L. A.; Moz Ruiz, D.; Afroz, S.; Mokraoui, I.; Velhal, P.; Morreale, P.; Sarma, A.; and Burnett, M. 2026. “Fast, easy, simple”? SES-diverse transfer ...

2026 arXiv
[2024]

In Al-Onaizan, Y .; Bansal, M.; and Chen, Y .-N., eds.,Proceedings of the 2024 Con- ference on Empirical Methods in Natural Language Pro- cessing, 6208–6226

Evaluating LLMs for Targeted Concept Simplifica- tion for Domain-Specific Texts. In Al-Onaizan, Y .; Bansal, M.; and Chen, Y .-N., eds.,Proceedings of the 2024 Con- ference on Empirical Methods in Natural Language Pro- cessing, 6208–6226. Miami, Florida, USA: Association for C...

2024
[2026]

Haver, H

Beyond Fixed Psychological Personas: State Beats Trait, but Language Models are State-Blind.arXiv preprint arXiv:2601.15395. Haver, H. L.; Lin, C. T.; Sirajuddin, A.; Yi, P. H.; and Jeudy, J. 2023. Use of ChatGPT, GPT-4, and Bard to Improve Readability of ChatGPT’s Answers to ...

2023 arXiv

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [12]

Learning Style: Tinkering-Oriented You are a programmer’s assistant

If I’m going to use a new feature or technology, I use very clear directions or help from someone else to learn it. Learning Style: Tinkering-Oriented You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and should follow eth- ical guidelines and promo...

[2] [13]

I enjoy finding the lesser-known features and capabilities of the devices and software I use

[3] [14]

I only look to instruction manuals as a last resort

I don’t follow instruction manuals. I only look to instruction manuals as a last resort

[4] [15]

I’m never satisfied with the default settings for my devices; I customize them in some way

[5] [16]

My first step in learning new technology is experimenting and tinkering with it

[6] [17]

I explore areas of a new application or service before it is time for me to use it

[7] [18]

to learn new software

I don’t need guidance, as in booklets, video how-tos sugges- tions, etc. to learn new software. Additionally, your user is more likely to align with the follow- ing statement:

[8] [19]

15 Self-Efficacy: Lower You are a programmer’s assistant

In order to learn new technology, I tinker with it, constructing my own understanding of how it works. 15 Self-Efficacy: Lower You are a programmer’s assistant. You can answer concep- tual programming questions and explain what code samples do. Your responses are helpful and harmless and should follow eth- ical guidelines and promote positive behavior. Yo...