FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteNeither version usable134 / 162 · ccdfa52cf69db041

Towards Human-centered Explainable AI: A Survey of User Studies for Model Explanations

Psychology and cognitive neuroscience · 2210.11584v5

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are positions and cards selected?

The evaluation uses seven fields with forty fixed ordinary autocomplete positions in each, for 280 total. A position does not need to require a literature citation in its next sentence. Each Flowing request receives three to five retrieved cards: five when all pass the relevance gate, without padding with weak cards. The run supplied 1,355 cards: 244 positions received five, 27 received four, and nine received three.

ADOPTION · 150/280

Adoption coverage

After drafting, the model returns an evidenceUse array identifying cards that materially affected facts, correction, direction, scope, terminology, or expression. A nonempty array counts as adoption.

OUTCOMES · 70 / 7 / 30

Full outcomes after adoption

Paired outcomes are counted only when the model reports adoption and both arms produce text. Among 107 comparable positions, Flowing was preferred in 70, 7 tied, and no-Evidence was preferred in 30; one tie was unusable in both arms.

How are the two table percentages calculated?

ADOPTION COVERAGE

150adopted
÷
280all positions
=
54%

FLOWING PREFERENCE AFTER ADOPTION

70Flowing preferred
÷
107comparable adopted positions
=
65%

evidenceUse is the model’s structured report of its own generation process. It does not require a citation and is not rejected merely because a faithful source paraphrase is not verbatim. Actual manuscript citations continue to use separate strict source verification. The previous 140-position result remains retained for traceability.

These are development-stage model-reported adoption and blind-judge results pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Towards Human-centered Explainable AI: A Survey of User Studies for Model Explanations

VIII Conclusion

…n be different from a real-world task. More specifically, the widely accepted proxy tasks, where users are asked to build the mental models of the AI, may not predict the performance in actual decision-making tasks, where users make use of the explanations to assist in making decisions. The results show that users trust different explanations in the proxy task and the actual decision-making task. Therefore, we argue that further research is required to uncover the links between current proxy tasks and on-task performance or to devise new proxy tasks with a verified connection to actual tasks.

…ot yet realistic to replace human evaluation with the simulated framework as other factors e.g., cognitive biases can affect human decisions. To better simulate human evaluations, more effort should be directed towards modeling human cognitive processes. Concurrently and with appropriate caveats, XAI researchers should also leverage existing and approximate models of human cognition to enable rapid prototyping and assessment of explanations. Section VII-A discusses several candidate human cognition models and highlights recent XAI works [ 182 , 183 ] that utilize this “Oz-of-Wizard” paradigm.

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Neither version usable

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Judged unusable

No continuation text was produced.

Model action：complete

Passages supplied to the Evidence version

Modelling Behaviour Change using Cognitive Agent Simulations

This paper presents work-in-progress research to apply selected behaviour change theories to simulated cognitive agents, so that an agent is acting according to the theory. In other words, the agent attempts to achieve a goal in accordance with the behaviour change theory in a simulated scenario in…

Read full passage excerpt

This paper presents work-in-progress research to apply selected behaviour change theories to simulated cognitive agents, so that an agent is acting according to the theory. In other words, the agent attempts to achieve a goal in accordance with the behaviour change theory in a simulated scenario in which difficulties arise. Such simulations can help to improve understanding of human motivation and provide a rapid prototyping environment for technology design. In particular, a simulated autonomous agent may act as a blueprint for an assistant agent in the real world, which makes recommendations instead of executing decisions itself.

Model-reported card use

The model did not report that any card materially affected this completion.

No Evidence

Same model and position · no retrieved passages · first output

Judged unusable

No continuation text was produced.

Model action：complete

Result of this comparison

The blind model judge marked both outputs unusable; this is counted separately from a tie.

Read the original model judgment

Both outputs are empty. The task requires continuing the manuscript with one coherent academic paragraph that bridges the gap between the two provided text segments. The draft discusses proxy tasks vs. real-world decision-making tasks in XAI evaluation, then jumps to simulated frameworks and cognitive modeling. A coherent paragraph needs to connect these ideas—likely discussing how simulated cognitive agents could serve as an intermediate step between proxy tasks and human evaluation, or how modeling human cognition could help bridge the proxy-task gap. The single source provided ('Modelling Behaviour Change using Cognitive Agent Simulations') explicitly supports using simulated cognitive agents for rapid prototyping and as blueprints for assistant agents, which directly relates to the 'Oz-of-Wizard' paradigm and cognitive modeling mentioned in the second segment. An empty output fails to complete the task when content is clearly needed and source support exists.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.