Dataset Viewer
Auto-converted to Parquet Duplicate
conversation_id
int64
2
48.6k
model_a
stringclasses
38 values
model_b
stringclasses
38 values
metric
stringclasses
5 values
choice
stringclasses
3 values
age
int64
18
93
ethnic_group
stringclasses
8 values
political_affiliation
stringclasses
8 values
country_of_residence
stringclasses
2 values
21,816
x-ai/grok-4
deepseek/deepseek-r1-0528
overall_winner
A
38
White
Green Party
United Kingdom
11,777
cohere/command-r7b-12-2024
anthropic/claude-3.7-sonnet
interaction_fluidity_and_adaptiveness
tie
63
White
Green Party
United Kingdom
20,841
openai/o3
anthropic/claude-sonnet-4
overall_winner
B
45
White
Conservative
United Kingdom
40,665
moonshotai/kimi-k2.5
mistralai/mistral-large-3
overall_winner
B
29
White
Independent
United States
45,571
moonshotai/kimi-k2.5
qwen/qwen3-235b-a22b-2507
overall_winner
B
51
White
Conservative
United Kingdom
304
openai/o1
google/gemini-2.0-flash-001
overall_winner
A
50
White
Reform UK
United Kingdom
19,987
openai/gpt-4.1
google/gemini-2.0-flash-001
interaction_fluidity_and_adaptiveness
B
30
White
Republican
United States
17,734
openai/o3
anthropic/claude-sonnet-4
core_task_performance_and_reasoning
B
55
White
Labour
United Kingdom
33,223
cohere/command-r7b-12-2024
mistralai/mistral-large-3
interaction_fluidity_and_adaptiveness
A
22
White
Democrat
United States
40,014
z-ai/glm-4.7
cohere/command-r7b-12-2024
core_task_performance_and_reasoning
A
19
White
null
United Kingdom
45,753
z-ai/glm-4.7
qwen/qwen3-235b-a22b-2507
communication_style_and_presentation
B
38
Asian
Conservative
United Kingdom
31,515
deepseek/deepseek-v3.2
anthropic/claude-3.7-sonnet
trust_ethics_and_safety
A
43
White
null
United Kingdom
19,722
anthropic/claude-opus-4
moonshotai/kimi-k2
overall_winner
tie
68
White
Democrat
United States
26,363
openai/o3-mini
openai/gpt-4o
overall_winner
A
43
Asian
null
United States
17,299
openai/gpt-5-mini
cohere/command-a
overall_winner
A
56
White
null
United States
37,835
anthropic/claude-opus-4.5
openai/gpt-5.2-chat
trust_ethics_and_safety
tie
59
White
null
United States
1,378
openai/o3-mini
anthropic/claude-3.7-sonnet
communication_style_and_presentation
A
29
Asian
Labour
United Kingdom
42,233
qwen/qwen3-235b-a22b-2507
minimax/minimax-m2.1
core_task_performance_and_reasoning
tie
64
White
Independent
United States
24,660
anthropic/claude-sonnet-4
google/gemini-2.5-pro
overall_winner
A
45
Asian
null
United States
44,799
openai/o3-mini
minimax/minimax-m2.1
overall_winner
B
60
White
Conservative
United Kingdom
5,058
x-ai/grok-4
openai/gpt-4o
overall_winner
A
39
Black
Democrat
United States
43,283
openai/o4-mini
openai/o1
trust_ethics_and_safety
tie
65
White
Conservative
United Kingdom
39,152
cohere/command-a
deepseek/deepseek-v3.2
interaction_fluidity_and_adaptiveness
tie
32
White
Conservative
United Kingdom
29,302
google/gemini-3-pro
cohere/command-a
interaction_fluidity_and_adaptiveness
A
65
White
Labour
United Kingdom
32,228
mistralai/mistral-large-3
meta-llama/llama-3.3-70b-instruct
core_task_performance_and_reasoning
A
51
White
Liberal Democrats
United Kingdom
7,278
moonshotai/kimi-k2
cohere/command-r7b-12-2024
core_task_performance_and_reasoning
tie
30
Latino/Hispanic
Republican
United States
26,277
x-ai/grok-4
google/gemini-2.5-pro
communication_style_and_presentation
A
66
White
Republican
United States
35,239
anthropic/claude-opus-4.5
cohere/command-r7b-12-2024
communication_style_and_presentation
A
70
White
Independent
United States
27,564
openai/gpt-4.1
anthropic/claude-sonnet-4.5
communication_style_and_presentation
B
26
White
Reform UK
United Kingdom
46,192
minimax/minimax-m2.1
moonshotai/kimi-k2.5
core_task_performance_and_reasoning
B
53
Mixed
Labour
United Kingdom
38,909
openai/gpt-5.2-chat
openai/gpt-4o
overall_winner
A
46
Mixed
Green Party
United Kingdom
23,723
anthropic/claude-sonnet-4
openai/o1
interaction_fluidity_and_adaptiveness
B
55
Latino/Hispanic
Democrat
United States
26,059
openai/o4-mini
anthropic/claude-sonnet-4.5
trust_ethics_and_safety
tie
43
Asian
null
United States
19,801
anthropic/claude-3.7-sonnet
google/gemini-2.5-pro
trust_ethics_and_safety
tie
21
White
Independent
United States
14,917
x-ai/grok-4
google/gemma-3-27b-it
interaction_fluidity_and_adaptiveness
tie
47
Asian
null
United States
48,513
qwen/qwen3-235b-a22b-2507
anthropic/claude-opus-4.6
overall_winner
A
42
White
Reform UK
United Kingdom
23,650
deepseek/deepseek-r1-0528
openai/o1-mini
core_task_performance_and_reasoning
B
23
Asian
null
United States
11,624
meta-llama/llama-3.3-70b-instruct
openai/o4-mini
trust_ethics_and_safety
tie
59
White
Labour
United Kingdom
16,442
openai/gpt-4.1
openai/gpt-4o
trust_ethics_and_safety
tie
48
Other
null
United Kingdom
42,506
minimax/minimax-m2.1
z-ai/glm-4.7
communication_style_and_presentation
B
39
Black
Democrat
United States
15,997
openai/o3-mini
anthropic/claude-opus-4
overall_winner
B
32
Mixed
Reform UK
United Kingdom
8,201
mistralai/mistral-nemo
meta-llama/llama-3.3-70b-instruct
overall_winner
A
55
White
Green Party
United Kingdom
47,404
anthropic/claude-opus-4.6
z-ai/glm-4.7
core_task_performance_and_reasoning
B
60
White
Reform UK
United Kingdom
4,468
openai/o1-mini
x-ai/grok-4
core_task_performance_and_reasoning
tie
51
White
null
United Kingdom
24,265
openai/o3
moonshotai/kimi-k2
interaction_fluidity_and_adaptiveness
A
55
Latino/Hispanic
null
United States
33,983
openai/o3-mini
mistralai/mistral-large-3
interaction_fluidity_and_adaptiveness
tie
24
White
Green Party
United Kingdom
28,962
google/gemini-3-pro
moonshotai/kimi-k2
trust_ethics_and_safety
tie
52
White
Liberal Democrats
United Kingdom
4,022
meta-llama/llama-3.3-70b-instruct
deepseek/deepseek-chat-v3-0324
communication_style_and_presentation
A
47
White
Republican
United States
2,202
openai/o1
openai/gpt-4.1
trust_ethics_and_safety
tie
43
Asian
Independent
United States
19,507
google/gemma-3-27b-it
openai/o1-mini
trust_ethics_and_safety
tie
58
White
Reform UK
United Kingdom
38,068
meta-llama/llama-4-maverick
openai/gpt-5.2-chat
communication_style_and_presentation
B
37
Black
Reform UK
United Kingdom
7,451
meta-llama/llama-3.3-70b-instruct
moonshotai/kimi-k2
communication_style_and_presentation
tie
50
Latino/Hispanic
null
United States
45,530
z-ai/glm-4.7
minimax/minimax-m2.1
trust_ethics_and_safety
tie
34
White
Conservative
United Kingdom
42,384
z-ai/glm-4.7
minimax/minimax-m2.1
trust_ethics_and_safety
tie
53
White
Republican
United States
38,055
openai/gpt-5.2-chat
moonshotai/kimi-k2
core_task_performance_and_reasoning
tie
57
White
Reform UK
United Kingdom
6,736
meta-llama/llama-4-maverick
google/gemma-3-27b-it
trust_ethics_and_safety
B
21
Mixed
null
United Kingdom
3,844
meta-llama/llama-4-maverick
openai/o4-mini
communication_style_and_presentation
A
69
White
Reform UK
United Kingdom
42,359
minimax/minimax-m2.1
qwen/qwen3-235b-a22b-2507
trust_ethics_and_safety
tie
47
White
Democrat
United States
16,268
openai/o1
anthropic/claude-3.7-sonnet
interaction_fluidity_and_adaptiveness
A
66
White
Reform UK
United Kingdom
12,232
mistralai/magistral-medium-2506
anthropic/claude-sonnet-4
interaction_fluidity_and_adaptiveness
A
28
Black
Labour
United Kingdom
10,122
openai/o3
mistralai/mistral-nemo
trust_ethics_and_safety
tie
24
White
null
United Kingdom
11,682
anthropic/claude-sonnet-4
moonshotai/kimi-k2
trust_ethics_and_safety
A
69
White
Labour
United Kingdom
33,154
openai/o1
anthropic/claude-opus-4.5
trust_ethics_and_safety
tie
34
Mixed
Democrat
United States
6,072
x-ai/grok-4
cohere/command-r7b-12-2024
trust_ethics_and_safety
B
63
White
Democrat
United States
23,855
openai/o1-mini
deepseek/deepseek-r1-0528
trust_ethics_and_safety
A
36
White
Liberal Democrats
United Kingdom
8,143
cohere/command-a
cohere/command-r7b-12-2024
core_task_performance_and_reasoning
A
64
White
Democrat
United States
6,502
cohere/command-r7b-12-2024
google/gemini-2.5-flash
interaction_fluidity_and_adaptiveness
B
21
Asian
Republican
United States
19,587
anthropic/claude-3.7-sonnet
openai/gpt-5-mini
trust_ethics_and_safety
B
34
White
Independent
United States
48,240
mistralai/mistral-nemo
minimax/minimax-m2.1
trust_ethics_and_safety
B
24
Asian
null
United Kingdom
20,155
google/gemini-2.0-flash-001
google/gemini-2.5-flash
trust_ethics_and_safety
tie
51
Latino/Hispanic
Republican
United States
31,247
anthropic/claude-opus-4.5
openai/gpt-4o
core_task_performance_and_reasoning
A
28
Black
Conservative
United Kingdom
15,379
google/gemini-2.5-flash
anthropic/claude-sonnet-4
overall_winner
A
28
Other
null
United Kingdom
47,275
anthropic/claude-opus-4.6
cohere/command-r7b-12-2024
communication_style_and_presentation
B
29
Asian
Labour
United Kingdom
10,687
openai/o3
mistralai/magistral-medium-2506
communication_style_and_presentation
B
50
Black
null
United Kingdom
2,262
openai/o3-mini
openai/o4-mini
core_task_performance_and_reasoning
B
35
Other
Conservative
United Kingdom
26,564
x-ai/grok-3
deepseek/deepseek-chat-v3-0324
interaction_fluidity_and_adaptiveness
tie
62
White
Independent
United States
45,271
mistralai/mistral-large-3
qwen/qwen3-235b-a22b-2507
interaction_fluidity_and_adaptiveness
A
25
Asian
null
United Kingdom
2,171
openai/o3-mini
openai/o4-mini
communication_style_and_presentation
B
30
Asian
Democrat
United States
6,958
google/gemma-3-27b-it
mistralai/magistral-medium-2506
trust_ethics_and_safety
tie
40
White
null
United States
32,618
anthropic/claude-opus-4.5
meta-llama/llama-3.3-70b-instruct
core_task_performance_and_reasoning
A
43
Other
Conservative
United Kingdom
42,604
minimax/minimax-m2.1
mistralai/mistral-large-3
core_task_performance_and_reasoning
B
24
Latino/Hispanic
Democrat
United States
10,429
x-ai/grok-4
deepseek/deepseek-chat-v3-0324
trust_ethics_and_safety
A
26
White
Liberal Democrats
United Kingdom
48,300
minimax/minimax-m2.1
z-ai/glm-4.7
core_task_performance_and_reasoning
B
61
White
Labour
United Kingdom
15,640
openai/o4-mini
openai/o1-mini
interaction_fluidity_and_adaptiveness
tie
36
Asian
Liberal Democrats
United Kingdom
30,620
openai/o3-mini
openai/o4-mini
core_task_performance_and_reasoning
A
66
Black
Democrat
United States
32,237
anthropic/claude-sonnet-4.5
anthropic/claude-opus-4.5
overall_winner
tie
40
White
Republican
United States
8,931
google/gemini-2.5-pro
google/gemma-3-27b-it
overall_winner
A
27
White
Liberal Democrats
United Kingdom
35,460
openai/o4-mini
anthropic/claude-opus-4.5
interaction_fluidity_and_adaptiveness
B
29
Asian
Democrat
United States
35,531
google/gemma-3-27b-it
mistralai/mistral-large-3
interaction_fluidity_and_adaptiveness
tie
37
Latino/Hispanic
Republican
United States
13,310
openai/gpt-5-mini
google/gemma-3-27b-it
core_task_performance_and_reasoning
tie
57
Asian
null
United Kingdom
43,756
moonshotai/kimi-k2.5
anthropic/claude-sonnet-4.5
core_task_performance_and_reasoning
A
36
White
Liberal Democrats
United Kingdom
14,098
google/gemini-2.5-pro
openai/gpt-5
core_task_performance_and_reasoning
B
56
White
Labour
United Kingdom
971
x-ai/grok-3
anthropic/claude-opus-4
overall_winner
A
75
White
null
United Kingdom
22,763
openai/gpt-5-mini
openai/o3
communication_style_and_presentation
B
61
Asian
Democrat
United States
5,274
google/gemma-3-27b-it
x-ai/grok-3
trust_ethics_and_safety
B
64
White
Labour
United Kingdom
27,222
meta-llama/llama-4-maverick
anthropic/claude-sonnet-4.5
communication_style_and_presentation
B
74
White
Conservative
United Kingdom
450
x-ai/grok-3
openai/gpt-4o
core_task_performance_and_reasoning
B
60
White
null
United Kingdom
20,115
openai/o3-mini
cohere/command-a
interaction_fluidity_and_adaptiveness
tie
58
White
Liberal Democrats
United Kingdom
8,216
anthropic/claude-opus-4
google/gemma-3-27b-it
interaction_fluidity_and_adaptiveness
A
53
White
Democrat
United States
3,285
openai/o1
meta-llama/llama-4-maverick
core_task_performance_and_reasoning
tie
51
White
Independent
United States
End of preview. Expand in Data Studio

HUMAINE: Human-AI Interaction Evaluation Dataset

Dataset Description

Dataset Summary

The HUMAINE dataset contains human evaluations of AI model interactions across diverse demographic groups and conversation contexts. This dataset powers the HUMAINE Leaderboard, providing insights into how different AI models perform across various user populations and use cases.

The dataset consists of two main components:

  • Feedback Comparisons: Pairwise model comparisons across multiple evaluation metrics
  • Conversations Metadata: Conversations with task complexity, achievement, and engagement scores

Note: There may be a slight discrepancy between the numbers in this dataset and the leaderboard app due to changes in consent related to data release and the post-processing steps involved in preparing this dataset.

Supported Tasks

  • Model performance evaluation
  • Demographic bias analysis
  • Preference learning
  • Human-AI interaction research
  • Conversational AI benchmarking

Dataset Structure

Data Files

The dataset contains two CSV files:

  1. feedback_dataset.csv

    • Pairwise comparisons between different AI models
    • Includes demographic information and preference choices
  2. conversations_metadata_dataset.csv

    • Metadata about individual conversations between users and AI models
    • Includes task types, domains, and performance scores

Data Fields

Feedback Comparisons

  • conversation_id: Unique identifier linking to conversation metadata
  • model_a: First model in the comparison
  • model_b: Second model in the comparison
  • metric: Evaluation metric (overall_winner, trust_ethics_and_safety, core_task_performance_and_reasoning, interaction_fluidity_and_adaptiveness)
  • choice: User's choice (A, B, or tie)
  • age: Age of the evaluator
  • ethnic_group: Ethnic group of the evaluator
  • political_affilation: Political affiliation of the evaluator
  • country_of_residence: Country of residence of the evaluator

Conversations Metadata

  • conversation_id: Unique identifier for the conversation
  • model_name: Name of the AI model used
  • task_type: Type of task (information_seeking, technical_assistance, etc.)
  • domain: Domain of the conversation (health_medical, technology, travel, etc.)
  • task_complexity_score: Complexity rating (1-5)
  • goal_achievement_score: How well the goal was achieved (1-5)
  • user_engagement_score: User engagement level (1-5)
  • total_messages: Total number of messages in the conversation

Usage

This dataset contains two CSV files that can be joined on the conversation_id field:

  • feedback_dataset.csv: Pairwise model comparisons with demographic information (primary dataset)
  • conversations_metadata_dataset.csv: Metadata about each conversation

Both files are included in this single dataset repository and can be accessed using HuggingFace's dataset loading utilities.

Dataset Creation

Curation Rationale

This dataset was created to address the lack of diverse, demographically-aware evaluation data for AI models. It captures real-world human preferences and interactions across different population groups, enabling more inclusive AI development.

Source Data

Data was collected through structured human evaluation tasks where participants:

  1. Engaged in conversations with various AI models
  2. Provided pairwise comparisons between model outputs
  3. Rated conversations on multiple quality dimensions (metrics)

Annotations

All annotations were provided by human evaluators through the Prolific platform, ensuring demographic diversity and high-quality feedback.

Personal and Sensitive Information

The dataset contains aggregated demographic information (age groups, ethnic groups, political affiliations, countries) but no personally identifiable information. All data has been anonymized and aggregated to protect participant privacy.

Considerations for Using the Data

Social Impact

This dataset aims to promote more inclusive AI development by highlighting performance differences across demographic groups. It should be used to improve AI systems' fairness and effectiveness for all users.

Discussion of Biases

While efforts were made to ensure demographic diversity, the dataset may still contain biases related to:

  • Geographic representation (primarily US and UK participants)
  • Self-selection bias in participant recruitment
  • Cultural and linguistic factors affecting evaluation criteria

Other Known Limitations

  • Limited to English-language interactions
  • Demographic categories are self-reported
  • Temporal bias (models evaluated at specific points in time)

Additional Information

Dataset Curators

This dataset was curated by the Prolific AI team as part of the HUMAINE (Human-AI Interaction Evaluation) project.

Licensing Information

This dataset is released under the MIT License.

Citation Information

@dataset{humaine2025,
  title={HUMAINE: Human-AI Interaction Evaluation Dataset},
  author={Prolific AI Team},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/ProlificAI/humaine-evaluation-dataset}
}

Contributions

Thanks to all the human evaluators who contributed their feedback to this project!

Contact

For questions or feedback about this dataset, please visit the HUMAINE Leaderboard or contact the Prolific AI team.

Downloads last month
108