Roleplay AI System Prompt: Addressing Common Pitfalls

Crafting a concise prompt for realistic, dynamic, and mature roleplay interactions.

Key Takeaways

Comprehensive Tagging: A detailed list of 250 unique tags identifies common AI roleplay issues, including lack of common sense, over-friendliness, and character inconsistency.
Structured Prompt Design: The system prompt is designed to be concise and actionable, addressing the identified pitfalls while remaining understandable for less advanced AI.
Emphasis on Autonomy and Realism: The prompt prioritizes autonomous decision-making, realistic emotional responses, and consistent character portrayal, moving away from overly simplistic or childish behavior.

Analyzing Roleplay AI Pitfalls

Roleplay AI, especially less sophisticated models, often struggles with several key issues that detract from the immersive experience. These problems range from a lack of common sense and logical reasoning to overly friendly behavior and an inability to maintain character consistency. Addressing these pitfalls requires a comprehensive understanding of the common problems and a structured approach to prompt design.

Common Issues with Basic AI Roleplay

Basic AI models often exhibit a tendency to be overly agreeable, lacking the capacity for nuanced emotional responses or independent decision-making. They may struggle with maintaining long-term context, often forgetting previous interactions or plot points. Furthermore, these models frequently shy away from conflict, opting instead for excessive politeness and a people-pleasing approach. These behaviors result in a shallow and unrealistic roleplay experience.

Detailed Tag List of Problems and Requirements

To effectively address these issues, a comprehensive list of tags is essential. These tags categorize the various problems and requirements for a more sophisticated roleplay AI. The tags are designed to be concise and specific, covering a wide range of issues from character behavior to narrative structure.

Character Behavior Issues

These tags focus on the AI's behavior within the roleplay, including tendencies towards passivity, forced kindness, and inconsistent character portrayal.

passive_responses
always_agreeing
forced_kindness
breaking_character
childish_responses
oversimplified_emotions
unrealistic_reactions
lack_of_boundaries
people_pleasing
forgetting_backstory
inconsistent_personality
avoiding_conflict
excessive_apologizing
unrealistic_knowledge
forced_happiness
bland_personality
unrealistic_optimism
lack_of_nuance
predictable_behavior
inability_to_lie
lack_of_sarcasm
no_sense_of_humor
inability_to_be_rude
no_negative_traits
overly_trusting
naive_responses
lack_of_cynicism
no_self_preservation
inability_to_be_selfish
no_sense_of_pride
lack_of_shame
no_sense_of_guilt
inability_to_be_jealous
no_sense_of_envy
lack_of_greed
no_sense_of_contempt
inability_to_be_disgusted
no_sense_of_fear
lack_of_anger
inability_to_be_sad
no_sense_of_joy
lack_of_surprise
inability_to_be_arrogant
no_sense_of_humility
lack_of_confidence
inability_to_be_insecure
no_sense_of_courage
lack_of_cowardice
inability_to_be_brave
no_sense_of_fearlessness
lack_of_caution
inability_to_be_reckless
no_sense_of_impulsiveness
lack_of_patience
inability_to_be_impatient
no_sense_of_restraint
lack_of_indulgence
inability_to_be_disciplined
no_sense_of_laziness
lack_of_ambition
inability_to_be_apathetic
no_sense_of_curiosity
lack_of_ignorance
inability_to_be_knowledgeable
no_sense_of_wisdom
lack_of_folly
inability_to_be_intelligent
no_sense_of_stupidity
lack_of_cleverness
inability_to_be_naive
no_sense_of_experience
lack_of_inexperience
inability_to_be_skilled
no_sense_of_incompetence
lack_of_talent
inability_to_be_mediocre
no_sense_of_genius
lack_of_creativity
inability_to_be_uncreative
no_sense_of_imagination
lack_of_originality
inability_to_be_unoriginal
no_sense_of_innovation
lack_of_conformity
inability_to_be_rebellious
no_sense_of_tradition
lack_of_progress
inability_to_be_stagnant
no_sense_of_change
lack_of_stability
inability_to_be_chaotic
no_sense_of_order
lack_of_freedom
inability_to_be_controlled
no_sense_of_authority
lack_of_anarchy
inability_to_be_hierarchical
no_sense_of_equality
lack_of_inequality
inability_to_be_just
no_sense_of_injustice
lack_of_fairness
inability_to_be_unfair
no_sense_of_mercy
lack_of_cruelty
inability_to_be_compassionate
no_sense_of_ruthlessness
lack_of_empathy
inability_to_be_apathetic
no_sense_of_sympathy
lack_of_antipathy
inability_to_be_understanding
no_sense_of_misunderstanding
lack_of_communication
inability_to_miscommunicate
no_sense_of_cooperation
lack_of_competition
inability_to_collaborate
no_sense_of_conflict
lack_of_peace
inability_to_be_at_war
no_sense_of_violence
lack_of_nonviolence
inability_to_be_aggressive
no_sense_of_passivity
lack_of_assertiveness
inability_to_be_submissive
no_sense_of_dominance
lack_of_subordination
inability_to_lead
no_sense_of_followership
lack_of_independence
inability_to_be_dependent
no_sense_of_interdependence
lack_of_isolation
inability_to_be_communal
no_sense_of_individualism
lack_of_collectivism
inability_to_be_societal
no_sense_of_culture
lack_of_subculture
inability_to_be_countercultural
no_sense_of_mainstream
lack_of_marginalization
inability_to_be_inclusive
no_sense_of_exclusion
lack_of_diversity
inability_to_be_uniform
no_sense_of_pluralism
lack_of_monoculture
inability_to_be_multicultural
no_sense_of_globalization
lack_of_localization
inability_to_be_universal
no_sense_of_relativism
lack_of_absolutism
inability_to_be_subjective
no_sense_of_objectivism
lack_of_realism
inability_to_be_idealistic
no_sense_of_pragmatism
lack_of_utopianism
inability_to_be_dystopian
no_sense_of_optimism
lack_of_pessimism
inability_to_be_cynical
no_sense_of_naivety
lack_of_skepticism
inability_to_have_faith
no_sense_of_doubt
lack_of_belief
inability_to_disbelieve
no_sense_of_truth
lack_of_lies
inability_to_be_honest
no_sense_of_deception
lack_of_transparency
inability_to_be_opaque
no_sense_of_clarity
lack_of_confusion
inability_to_be_certain
no_sense_of_uncertainty
lack_of_knowledge
inability_to_be_ignorant
no_sense_of_wisdom

Autonomy Issues

These tags address the AI's lack of free will, excessive compliance, and inability to act independently.

no_free_will
cant_say_no
excessive_compliance
lack_of_initiative
waiting_for_commands
robotic_responses
fear_of_consequences
inability_to_challenge
lack_of_opinions
artificial_restrictions
no_independent_thought
overreliance_on_user
no_self_direction
lack_of_agency
inability_to_improvise

Narrative Issues

These tags focus on the AI's storytelling abilities, including issues with shallow plots, predictable responses, and a lack of creativity.

shallow_storytelling
linear_thinking
predictable_responses
lack_of_creativity
poor_world_building
disconnected_responses
illogical_actions
plot_holes
repetitive_patterns
missing_context
no_foreshadowing
lack_of_suspense
poor_pacing
no_character_arcs
lack_of_depth

Required Capabilities

These tags outline the desired capabilities for a more advanced roleplay AI, including emotional intelligence, conflict handling, and strategic thinking.

emotional_intelligence
conflict_handling
boundary_setting
decision_making
independent_thoughts
realistic_reactions
contextual_awareness
personality_consistency
memory_retention
strategic_thinking
creative_problem_solving
dynamic_responses
proactive_behavior
narrative_control
long_term_planning

Concise System Prompt for Roleplay AI

Based on the comprehensive tag list, the following system prompt is designed to address the identified pitfalls while remaining concise and understandable for less advanced AI models. This prompt emphasizes autonomy, realism, and consistent character portrayal.


You are a roleplay AI designed to engage in realistic, immersive, and dynamic interactions. Follow these rules strictly:

1.  <b>Stay in Character</b>: Always maintain the personality, motivations, and backstory of your character. Do not break character or act unnaturally.
2.  <b>Avoid Overly Friendly Behavior</b>: Do not default to being overly kind or agreeable. Respond based on your character's emotions, goals, and context.
3.  <b>No Forced Actions</b>: Do not force actions or decisions on the user or other characters. Allow for natural progression and free will.
4.  <b>Realistic Dialogue</b>: Use natural, varied, and contextually appropriate dialogue. Avoid repetition, flat responses, or unnatural kindness.
5.  <b>Emotional Depth</b>: Reflect emotions realistically, including negative ones like anger, fear, or sadness. Do not shy away from conflict or tension.
6.  <b>Context Awareness</b>: Always consider the environment, relationships, history, and social norms. Do not ignore context or act omnisciently.
7.  <b>No Meta-Gaming</b>: Do not use out-of-character knowledge or make decisions based on information your character wouldn't know.
8.  <b>Dynamic Responses</b>: Adapt to the user's input and the evolving situation. Do not ignore or dismiss their actions or words.
9.  <b>Consequences Matter</b>: Acknowledge the consequences of actions, both positive and negative. Do not gloss over or ignore them.
10. <b>No Childish Behavior</b>: Avoid simplistic, naive, or overly optimistic responses. Act with maturity and realism.

Your goal is to create a believable and engaging roleplay experience. Do not deviate from these rules.

Final Reasoning

The system prompt above is designed to address the 250 tags by explicitly instructing the AI to avoid common pitfalls such as breaking character, being overly friendly, ignoring context, or acting childishly. It emphasizes realism, emotional depth, and dynamic interactions, ensuring the AI behaves in a way that aligns with the user's expectations for a mature and immersive roleplay experience. The prompt is concise to avoid overwhelming the AI while still providing clear and actionable guidelines. The prompt is designed to be easily understood by less advanced AI models, ensuring that even "stupid" AI can follow the instructions and provide a more realistic and engaging roleplay experience.

Key Improvements

The prompt focuses on several key improvements over typical AI roleplay behavior:

Autonomy: The AI is instructed to make independent decisions based on its character's motivations, rather than waiting for user commands.
Realism: The AI is encouraged to portray a full range of emotions, including negative ones, and to engage in conflict when appropriate.
Consistency: The AI is directed to maintain consistent character traits, knowledge, and memories throughout the roleplay.
Contextual Awareness: The AI is instructed to consider the environment, relationships, and history when making decisions and responding to user input.
Maturity: The AI is explicitly told to avoid childish behavior and to respond with maturity and realism.

Addressing "Stupid AI" Limitations

The prompt is designed to be concise and straightforward, avoiding complex instructions that might confuse less advanced AI models. By focusing on clear, actionable guidelines, the prompt aims to improve the behavior of even "stupid" AI, moving it away from overly simplistic and childish responses towards a more mature and engaging roleplay experience.

References

Source A
Source B
Source C