Fact-checked by Grok 2 weeks ago

AIML

Artificial Intelligence Markup Language (AIML) is an open-source, XML-based markup language designed for defining rules that govern the responses of natural language software agents, such as chatbots and virtual assistants. Developed by Richard S. Wallace in the mid-1990s as the foundational technology for the A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) chatbot, AIML structures knowledge through categories—each comprising a pattern to match user inputs and a template to generate outputs—enabling pattern-matching logic for conversational interactions. This rule-based approach facilitates straightforward implementation of dialogue systems without requiring advanced machine learning, supporting features like srai (recursive substitution for pattern generalization), think tags for internal processing, and extensibility via predicates for context-aware responses. AIML's simplicity and portability have made it a cornerstone for early chatbot development, powering platforms like Pandorabots and influencing open standards for multilingual, customizable agents deployable in customer service and educational tools. While effective for deterministic conversations, its reliance on predefined rules limits adaptability to novel inputs compared to modern probabilistic models, highlighting its role in pioneering accessible yet constrained natural language processing.

History

Origins and Development

Dr. Richard S. Wallace initiated the development of (AIML) in 1995 as part of the (A.L.I.C.E.) project, aiming to formalize rule-based for conversational agents in a structured markup format. This effort built upon the foundational pattern-matching techniques of Joseph Weizenbaum's program from 1966, extending scripted responses into a more scalable system for simulating human-like dialogue without requiring extensive programming expertise. Wallace's individual innovation focused on creating a tool that allowed developers to define behaviors through declarative rules, prioritizing simplicity and adaptability over brittle, hard-coded logic. AIML emerged as an XML-compliant specifically tailored for specifying input and corresponding output , enabling the construction of responsive agents that could handle varied queries via keyword and phrase recognition. Unlike prior ad-hoc scripting in early , AIML's structure emphasized human-readable categories—each containing a for user input matching and a for generating replies—facilitating easier maintenance and expansion of knowledge bases. This design choice reflected Wallace's goal of democratizing chatbot creation for non-experts, with initial implementations tested within the framework to produce coherent, context-aware interactions. By the late , AIML's core mechanism of hierarchical pattern prioritization—where more specific patterns override general ones to ensure precise and non-fragile responses—had been refined through iterative prototyping in , demonstrating robustness in handling ambiguous or off-topic inputs. Early deployments occurred in academic settings and among hobbyist programmers, where the language's rule-based approach proved effective for educational simulations and basic interfaces, laying groundwork for broader applications while highlighting limitations in deep semantic understanding. Wallace's solo efforts during this phase underscored AIML's origins in practical problem-solving for pattern-driven AI, distinct from statistical methods emerging elsewhere.

Standardization and Open-Sourcing

The (AIML) was formalized through collaborative efforts led by Richard S. Wallace between 1995 and 2002, during which the XML-based dialect evolved into AIML 1.0, establishing it as a for defining pattern-matching rules in conversational agents. This period saw the release of comprehensive and the core syntax, enabling developers worldwide to create and share behaviors without proprietary constraints. Wallace, through the non-profit ALICE A.I. Foundation, released AIML under open-source principles aligned with ideals, explicitly avoiding corporate control to promote unrestricted access and modification. This decision facilitated contributions from a global developer community, with AIML files and interpreters made freely available via repositories such as Code and , fostering iterative improvements in and response generation. In 2013, the Pandorabots community advanced the standard with the initial draft of AIML 2.0, incorporating enhancements like and to enable dynamic data handling and greater flexibility in bot logic without altering the foundational pattern-based architecture. These additions, refined through community review and updated in subsequent drafts by , expanded AIML's capacity for conditional processing and self-modification while maintaining with version 1.0. The open nature of these updates ensured decentralized evolution, with implementations verifiable through public specifications rather than vendor-locked ecosystems.

Key Milestones and Expansions

The , implemented using AIML, secured victories in the competition in 2000, 2001, and 2004, highlighting the markup language's capability to produce responses that judges rated as convincingly human-like during Turing Test-style interactions. These successes, achieved through precise and template-based generation, established AIML as a viable framework for advancing performance in constrained evaluation settings. By the early 2000s, AIML expanded beyond initial research prototypes via integrations into robust platforms, including Program D, a Java-based open-source interpreter that supported deployment of AIML bots with features like graphical interfaces and extensive testing for reliability. Similarly, Pandorabots emerged as a key commercial host for AIML scripts, enabling scalable bot creation and integrations for non-academic uses such as customer interaction prototypes. Further advancements materialized with AIML 2.0, ratified around 2018, which incorporated extensions from implementations like Pandorabots, adding capabilities such as wildcard expansions in conditions, richer media tags, and utility libraries for and string operations to enhance expressiveness without altering core syntax. Despite the ascendancy of data-driven large language models, AIML has persisted in post-2020 niches, particularly for lightweight applications in and mobile tools where low computational overhead and deterministic behavior suit resource-limited or privacy-sensitive deployments. Examples include LINE-based English learning chatbots leveraging AIML for contextual interaction and mobile AI psychologists employing it for simulations. Platforms like Pandorabots continue to support ongoing development, underscoring AIML's endurance for scenarios prioritizing interpretability over probabilistic scaling.

Technical Specifications

Core Syntax and Structure

Artificial Intelligence Markup Language (AIML) employs an XML-based declarative format where individual knowledge units, known as categories, define mappings from user inputs to bot responses without requiring probabilistic training or machine learning models. An AIML document typically begins with a root <aiml> element that encapsulates one or more <category> tags, each serving as a self-contained rule. This structure prioritizes human-readable, hand-crafted rules over automated data fitting, allowing for direct inspection and modification of the underlying logic. The <category> element pairs a <pattern> for specifying the input with a <template> for generating the output response. Patterns are expressed in uppercase normalized text to match user inputs after preprocessing steps like tokenization and , supporting exact phrase matching or generalized forms via wildcards. Wildcards include * and _, both matching one or more words in AIML 1.0, with * enabling capture for reuse in templates via <star/> elements, while _ prioritizes in matching hierarchies but does not capture. AIML extends this with ^ for zero or more words (non-capturing) and # for zero or one word, enhancing flexibility in pattern expressiveness without introducing . Patterns incorporate predicates—contextual variables such as or session data—for conditional matching, formatted as <pattern>BASE PATTERN <bot name="predicate"/> </pattern>. This allows rules to reference prior interactions or fixed bot attributes, fostering context-aware responses through explicit declarations rather than inferred states. Templates, conversely, consist of static text, AIML tags for dynamic insertion (e.g., <star index="1"/> to insert wildcard captures), or recursive elements like <srai> for response substitution. The <srai> tag redirects processing to another pattern by wrapping input text, enabling rule reuse and basic while preventing infinite loops through interpreter safeguards. This syntax's simplicity stems from its avoidance of statistical dependencies, relying instead on exhaustive rule authoring for deterministic behavior, which ensures full auditability as every output traces directly to authored categories. Developers author rules manually or semi-automatically from corpora, prioritizing over in large, unpredictable domains.

Pattern Matching and Response Generation

In AIML interpreters, user input undergoes prior to , which typically involves converting text to uppercase, stripping punctuation, expanding common contractions, and ensuring single spaces between words to standardize the input for consistent evaluation. This process, often implemented via substitution files or dedicated normalization tags, enables case-invariant matching against predefined patterns stored in the system's knowledge base. The core pattern matching employs a deterministic algorithm that prioritizes the longest and most specific match through a depth-first traversal of the Graphmaster, a trie-like directed graph structure where nodes represent words or wildcards from input, context (prior bot response), and topic predicates. Exact word matches receive highest priority, followed by prioritized wildcards (e.g., boundary-bound ones over general), ensuring that detailed patterns override broader ones; for instance, a multi-word pattern supersedes a single-word or wildcard equivalent. This graph-based storage facilitates efficient retrieval, with exact matches achieving constant-time O(1) lookups via internal hashing, even in datasets containing thousands of categories. Upon identifying a matching , the associated generates the response by evaluating embedded elements, such as those enabling internal state updates without visible output or selection from predefined alternatives to introduce controlled variability. If no exact or prioritized match occurs, the system falls back to the ultimate default category, typically a broad wildcard yielding a generic acknowledgment like "I don't understand," preventing silent failures. Implementations like Pandorabots (Java-based) and (Python) manage this dynamically, loading categories into for evaluation during interactions.

Categories and Advanced Features

The element constitutes the fundamental unit of AIML knowledge representation, pairing a for matching user inputs—via exact strings, wildcards such as * or _, or priority indicators like ^ and #—with a

References

  1. [1]
    AIML Foundation
    Artificial Intelligence Markup Language · Open. Not a black box. · Multilingual. AIML can be written in almost any natural language. · Flexible & Extensible.
  2. [2]
    (PDF) Artificial Intelligence MArkup Language: A Brief Tutorial
    Aug 5, 2025 · The purpose of this paper is to serve as a reference guide for the development of chatterbots implemented with the AIML language.
  3. [3]
    AIML Introduction - GeeksforGeeks
    Apr 28, 2025 · AIML (Artificial Intelligence Markup Language) is a description language used in the development of natural language software agents like chatbots and virtual ...Features of AIML · AIML Tags and their syntax · AIML Examples
  4. [4]
    AIML Fundamentals - Pandorabots Documentation
    AIML, or Artificial Intelligence Mark-up Language enables people to input knowledge into chatbots. If you prefer learning by video, check out our free AIML ...
  5. [5]
    What is AIML (Artificial Intelligence Markup Language)? - Infobip
    Oct 15, 2024 · AIML, or Artificial Intelligence Markup Language, is an XML dialect developers use to create natural language software agents commonly used in ...
  6. [6]
    Artificial Intelligence Markup Language (AIML) - Pandorabots
    The XML dialect called AIML was originally developed by Dr. Richard Wallace and a worldwide free software community between 1995 and 2002. AIML formed the basis ...Missing: origins project
  7. [7]
    The History of Artificial Intelligence - IBM
    1951. Marvin Minsky and Dean Edmunds build the first artificial neural network. The Stochastic Neural Analog Reinforcement Calculator (SNARC) is an early ...
  8. [8]
    The Story Of ELIZA: The AI That Fooled The World
    Sep 15, 2024 · 1995: ALICE (Artificial Linguistic Internet Computer Entity), developed by Richard Wallace, builds on ELIZA and uses a more sophisticated ...
  9. [9]
    The Anatomy of A.L.I.C.E. | SpringerLink
    This paper describes the history of A.L.I.C.E. and AIML-free software since 1995, noting that the theme and strategy of deception and pretense upon which AIML ...
  10. [10]
    [PDF] ARTIFICIAL INTELLIGENCE MARKUP LANGUAGE - arXiv
    ABSTRACT. The purpose of this paper is to serve as a reference guide for the development of chatterbots implemented with the AIML language.Missing: origins | Show results with:origins
  11. [11]
    Free ALICE AIML Set - Google Code
    May 21, 2011 · A free AIML set, in the English Language as spoken in the United States, authored by the AI Foundation, and the bot name is ALICE.
  12. [12]
    drwallace/aiml-en-us-foundation-alice - GitHub
    Free ALICE AIML. Contribute to drwallace/aiml-en-us-foundation-alice development by creating an account on GitHub.
  13. [13]
    What's new in AIML 2.0 - Bot Libre
    Aug 21, 2015 · The AIML 2.0 draft was released in 2013, and last updated in 2014. AIML is supported by many different chat bot programs and hosting services.
  14. [14]
    AIML Reference - Pandorabots
    This powerful introduced in AIML 2.0 allows users to actually teach the bot new information. Categories generated by the learn element are stored in memory ...
  15. [15]
    ALICE talks her way to victory in AI challenge - ZDNET
    Oct 16, 2001 · ALICE, the Artificial Linguistic Internet Computer Entity, has won the Loebner prize, but remains decidedly underwhelmed at the dubious ...
  16. [16]
    ALICE Wins Loebner 2004 Prize 2004 - Slashdot
    Sep 21, 2004 · alicebotmaster writes "The A.L.I.C.E. chatbot won the 2004 Loebner Prize contest for most human computer. The contest, held on September 19 ...<|separator|>
  17. [17]
    noelbush-xx/programd: AIML-based conversational bot server - GitHub
    Program D is the most widely used free ("open source") AIML bot platform in the world. It is the most feature-complete, best-tested implementation of the ...
  18. [18]
    Pandorabots: Home
    Learn the ins and outs of AIML and the Pandorabots IDE, and how to integrate your APIs and internal systems. Read the Docs. Blog. Check out our blog for feature ...
  19. [19]
    AIML Docs
    Jun 20, 2018 · The basic step of AIML pattern matching is to match one input sentence against the bot's set of AIML categories. Because inputs and responses ...Missing: prioritization | Show results with:prioritization
  20. [20]
    Introducing: AIML Rich Media Tags - pandorabots-blog - Medium
    Jun 2, 2018 · Pandorabots and AIML give you a lot of flexibility to create your own custom AIML ... AIML Extensions. For example, you might have an application ...Missing: expansions | Show results with:expansions
  21. [21]
    Investigation of the Influence of Artificial Intelligence Markup ...
    Apr 6, 2022 · This study is intended to create an innovative contextual English learning environment making use of the widely used communication software, LINE ChatBot.
  22. [22]
    Transforming healthcare with chatbots: Uses and applications ... - NIH
    Mar 17, 2025 · An AI-powered mobile chatbot psychologist is introduced, utilizing Artificial Intelligence Markup Language (AIML) and Cognitive Behavioral ...
  23. [23]
    Program AB - Google Code
    Jan 10, 2013 · Program AB is the reference implementation of the AIML 2.0 draft specification. AIML is a widely adopted standard for creating chat bots and ...
  24. [24]
    program-ab - ExtendingAIML.wiki - Google Code
    Program AB defines a Java Interface called AIMLProcessorExtension that you can use to define new AIML tags. A class implementing AIMLProcessorExtension must ...
  25. [25]
    AIML Overview - Pandorabots
    Wallace. AIML, or Artificial Intelligence Mark-up Language enables people to input knowledge into chat-bots based on the A.L.I.C.E free software technology.
  26. [26]
    [PDF] Chapter 00 The Anatomy of A.L.I.C.E. - FreeShell
    Abstract: This paper is a technical presentation of Artificial Linguistic Internet. Computer Entity (A.L.I.C.E.) and Artificial Intelligence Markup Language.Missing: origins | Show results with:origins
  27. [27]
    ALICE and friends. Artificial Intelligence Reloaded - Marcel Gagné
    Jan 17, 2012 · Dr. Wallace won the 2000, 2001, and 2004 Loebner prize for the most 'human' program; that would be the bronze since no one has yet claimed ...
  28. [28]
    AI Chatbots in Education: Challenges and Opportunities - MDPI
    In 1995, Richard Wallace developed a chatbot named Alice [17]. Alice's design consisted of pattern-matching algorithms that analyzed and compared the user's ...
  29. [29]
    The Pandorabots Story
    Feb 15, 2014 · Pandorabots operates a free website used to build and deploy virtual personalities - in the form of human-like chatbots.Background · Inception · More About Pandorabots
  30. [30]
    Pandorabots Review: Open-Source AI Chatbot for Developers Not ...
    Deploy on multiple channels: You can place your Pandorabot on both voice and messaging channels such as WhatsApp, Slack, Facebook Messenger, Twilio, LINE, Viber ...Missing: commercial | Show results with:commercial
  31. [31]
    Pandorabots: A platform review 2023
    Dec 25, 2023 · Pandorabots is a free open-source-based website enabling you to develop and publish chatbots on the web. So should you purchase Pandorabots?
  32. [32]
    Frequently Asked Questions - Pandorabots Documentation
    The AIML 2.0 specification introduces a number of new features to the language that dramatically improve the natural language processing power of chatbots. The ...
  33. [33]
    Why Pandorabots
    Pandorabots has been in business since 2008. Our visionary founders are “Original Gangsters” when it comes to chatbots, and the platform has survived several ...Market Tested · Drawbacks To Pure... · Drawbacks To Drag-And-Drop...
  34. [34]
    Bot Possession: A Customer Service Solution | by Pandorabots
    Oct 30, 2014 · Bot possession allows a single agent to monitor dozens of conversations with the chatbot, and step in to a conversation when the chatbot is ...Missing: commercial game NPCs<|separator|>
  35. [35]
    Pandorabots Documentation
    The Pandorabots platform supports an open standard scripting language called Artificial Intelligence Markup Language (AIML). What is AIML? AIML is XML-based and ...
  36. [36]
    Creating Better NPCs - Game Developer
    Mar 18, 2013 · The easiest method appears to be as a free web service provided by Pandora Bots. There are also several interpreters available in different ...
  37. [37]
  38. [38]
    Pandorabots - Meta-Guide.com
    Pandorabots is used in a variety of applications, including customer service, education, entertainment, and more. It is designed to be easy to use and integrate ...
  39. [39]
    AIML+: Enhancing AIML for the Educational Domain Through ...
    Jul 22, 2025 · Artificial Intelligence Markup Language (AIML) is a standard DS framework adopted by many educators since it requires minimal programming ...
  40. [40]
    [PDF] (AI & Machine learning ) - IJNRD
    Mar 3, 2024 · What is ML? ML ... These limitations have spurred research into hybrid approaches combining AIML with machine learning techniques for improved ...
  41. [41]
    Artificial Intelligence Enabled Mobile Chatbot Psychologist using ...
    The paper discusses the development and implementation of the mobile chatbot psychologist, detailing the AIML-based conversational engine and the incorporation ...
  42. [42]
    Machine Learning vs. Rule Based Systems in NLP - Medium
    Dec 26, 2017 · Overall, a rule-based system is good at capturing a specific language phenomenon: it will decode the linguistic relationships between words to ...
  43. [43]
    Rule-Based vs LLM Chatbots — What's the Difference and Which ...
    Apr 24, 2025 · So here's a simple breakdown of rule-based vs LLM-based chatbots, and what I learned from working on both. Rule-Based Chatbots (like Rasa).
  44. [44]
    Rule-Based vs. LLM-Based AI Agents: A Side-by-Side Comparison
    Feb 3, 2025 · Rule-based AI agents operate on predefined rules, ensuring predictable and transparent decision-making, while LLM-based AI agents leverage deep learning for ...
  45. [45]
    A Comparative Analysis of Rule-Based and Deep Learning Models ...
    Nov 7, 2024 · The Rule-Based Model's transparency and strong theoretical foundation allow us to attribute shortcomings in both the corpus and external dataset ...
  46. [46]
    Interpreting Black-Box Models: A Review on Explainable Artificial ...
    Aug 24, 2023 · A black-box model in XAI refers to a machine learning model that operates as an opaque system where the internal workings of the model are not ...
  47. [47]
    (PDF) Machine Learning vs. Rule-Based NLP in Lisp - ResearchGate
    May 23, 2025 · Our findings suggest that while ML methods offer superior performance in large-scale applications, rule-based systems implemented in Lisp retain ...
  48. [48]
    Learn how AI can improve daily workflows in the legal sector
    Aug 15, 2024 · Rules-based AI is especially helpful for straightforward tasks that follow clear guidelines. However, its dependence on preset rules can make it ...Missing: persist | Show results with:persist
  49. [49]
    Rule Based System Vs Machine Learning System - GeeksforGeeks
    Aug 6, 2025 · Rule-based systems use explicit rules, while machine learning systems learn from data. Rule-based systems are static, while machine learning ...
  50. [50]
    Transparent Vs. Black-Box AI, New Software Development
    Apr 7, 2022 · While black-box AI systems can achieve great levels of accuracy and efficiency, rule-based AI systems are more dependable and reliable.
  51. [51]
    What Is AIML? | Full Form, Uses, and Examples - GoGloby
    Artificial Intelligence courses: AIML introduces symbolic reasoning and rule-based dialogue. Natural Language Processing classes: Demonstrates early chatbot ...
  52. [52]
    How I Use AIML for Lightweight Chatbot Development with Python
    Jul 11, 2025 · How I Use AIML for Lightweight Chatbot Development with Python ... offline without relying on external APIs ✓ Handle structured, FAQ ...
  53. [53]
    A Systematic Review of Chatbot Development for Low-Resource ...
    Mar 28, 2025 · ... (AIML): is. one of the rule-based technique for low-resource languages,. enabling developers to create structured conversation flows by.
  54. [54]
    11 Best Open Source AI Chatbots to Use in 2025 - ControlHippo
    Aug 19, 2025 · ... choice for lightweight chatbots built on rule-based AI. Features. Analyze log files; Code addition for application-specific custom AIML tags ...
  55. [55]
    Ask Alice
    ### Summary of AIML Advantages for Controlled Settings
  56. [56]
    Exploring Types of Chatbots: an AI/ML perspective - LinkedIn
    Feb 25, 2025 · Predictability: because the responses are predetermined, rule-based chatbots are predictable and easy to test, ensuring consistent behavior ...
  57. [57]
    [PDF] Rule-based AI chatbot - IJIRT
    ADVANTAGES. • Predictability and Control: Rule-based chatbots follow predefined patterns, ensuring predictable and controlled responses. This makes them ...
  58. [58]
    (PDF) AIML+: Enhancing AIML for the Educational Domain Through ...
    Aug 21, 2025 · Artificial Intelligence Markup Language (AIML) is a standard DS framework adopted by many educators since it requires minimal programming skills ...
  59. [59]
    [PDF] A Literature Survey of Recent Advances in Chatbots - arXiv
    Jan 17, 2022 · Furthermore, pattern matching rules are brittle, highly domain specific, and do not transfer well from one problem to the other. 4.1.2.
  60. [60]
    [PDF] Can rule-based chatbots outperform neural models without pre ...
    Sep 20, 2021 · This paper addresses the limitations of rule- based and end-to-end neural chatbots with lit- tle training data. We compare an AIML-based.Missing: benchmarks | Show results with:benchmarks
  61. [61]
    Comparative Analysis of Classical and Neural Networks based ...
    Sep 17, 2025 · While rule-based relies on predefined templates and responses, a neural network based relies on deep learning models. Rule-based are preferable ...Missing: benchmarks | Show results with:benchmarks
  62. [62]
    Comparison of Rule-based Chat Bots with Different Machine ...
    Jun 12, 2025 · Comparison of Rule-based Chat Bots with Different Machine Learning Models ... Algorithm Inspection for Chatbot Performance Evaluation.
  63. [63]
    Technology | Alice chatbot wins for third time - BBC NEWS
    Sep 20, 2004 · It was judged to be chattiest bot out of the four finalists in the Loebner Prize for artificial intelligence held in New York on Sunday.
  64. [64]
    Loebner Prize - Meta-Guide.com
    ALICE won the Loebner prize (2003) competition three times in 2000, 2001, and 2004 … …
  65. [65]
    [PDF] Comparative Review of Open-Source Chatbot Languages: AIML ...
    Program AB, an implementation of AIML. 2.0, incorporates features such as Sets, Maps, wildcards, and the ability to connect to remote bots and web services ...<|separator|>
  66. [66]
    Why Deep Learning is Not a Good Fit For Chatbots - LinkedIn
    Jun 23, 2017 · Facebook, when they launched the chatbot platform, assumed that buttonizing conversations could solve part of the combinatory explosion problem ...
  67. [67]
    Why Chatbots Failed and the Future of Conversational AI
    Dec 7, 2021 · They didn't use machine-learning effectively. AI is meant to learn over time. The bot should be able to learn from past interactions and improve ...
  68. [68]
    [PDF] Hybrid Educational Chatbot Integrating AIML 2.0 and Ontology ...
    Jun 30, 2025 · Combining the two offers a hybrid solution: AIML can handle lightweight FAQs and content delivery, while ChatScript can manage deeper logic ...
  69. [69]
    Beyond ChatGPT: A Hybrid Chatbot Model for Reliable Educational ...
    Oct 2, 2024 · AIML File Generation: The Pandorabots playground was used to generate AIML files from the dataset, creating the initial chatbot prototype.
  70. [70]
    Pandorabots: Home
    Multilingual. Write AIML scripts in any natural language ; Context-aware. Smarter than your average decision tree ; Open Standards. No platform lock-in: own and ...
  71. [71]
    A survey on chatbots and large language models - ScienceDirect.com
    This section explores the current developments in AI based chatbots, describing their functionality and applications. It also provides a detailed discussion of ...
  72. [72]
    Overview of Chatbots with special emphasis on artificial intelligence ...
    These Chatbots leverage natural language processing (NLP) and machine learning (ML) algorithms to comprehend and respond to user inquiries conversationally and ...
  73. [73]
    Bias and ethics of AI systems applied in auditing - A systematic review
    This study aimed to investigate the sources of bias and risks posed by AI systems applied in auditing and the complex downstream interactions and effects they ...
  74. [74]
  75. [75]
    Rule-Based vs AI-Powered Chatbots: What Works in 2025
    Oct 18, 2025 · A rule-based chatbot works like a decision tree. It matches user inputs to keywords or buttons and delivers fixed responses. These bots handle ...Missing: combinatorial explosion<|separator|>