Getting Started with AI Tools

Overview of Some Common Platforms

Important Disclaimer: The following features, limitations, and potential uses shared for specific platforms were known to be true and accurate at the time of creation of this resource. Updates to these platforms occur regularly and may alter the accuracy of these statements. Use your best judgment and hands-on experience to verify abilities of each platform. These statements are only meant for guidance as you begin your AI journey.

General-Purpose Large Language Models (LLMs)

ChatGPT (OpenAI)

• Features: User-friendly interface with conversation memory, handles text generation across multiple domains, supports code generation and analysis, integrates with plugins for expanded capabilities, and offers tiered subscription models with GPT-3.5 and GPT-4 options
  

• Limitations: Knowledge cutoff dates (varies by model version), potential for factual errors or "hallucinations," limited reasoning with complex numerical data, cannot access real-time information without plugins, and may struggle with highly specialized CSD terminology
  

• Use cases in CSD: Creating customized therapy materials, generating clinical documentation templates, simplifying research articles for patient education, brainstorming therapy activities, and developing student assessment questions
  

Claude (Anthropic)

• Features: Exceptionally large context window (up to 100,000 tokens), nuanced reasoning capabilities, strong performance on complex writing tasks, thoughtful analysis of ethical considerations, and specialized ability to work with uploaded documents
  

• Limitations: Similar knowledge limitations as other LLMs, more conservative responses in clinical domains, fewer integration options compared to ChatGPT, and limited multimodal capabilities
  

• Use cases in CSD: Analyzing lengthy assessment reports, creating comprehensive clinical documentation, developing evidence-based treatment protocols, generating detailed patient education materials, and evaluating ethical scenarios for student training
  

Gemini (Google)

• Features: Multimodal capabilities allowing image analysis alongside text, strong performance on factual knowledge tasks, integration with Google search results (in some versions), and native support for charts and data visualization
  

• Limitations: Inconsistent performance across different prompt types, more restricted in generating certain types of content, limited customization options, and variable response quality
  

• Use cases in CSD: Analyzing visual therapy materials, generating visual supports for AAC, interpreting charts and graphs from assessment data, research assistance with visualization, and creating instructional materials with visual components
  

Copilot (Microsoft)

• Features: Deep integration with Microsoft 365 apps, contextual awareness of documents being edited, ability to summarize long documents, email drafting assistance, and meeting transcription capabilities
  

• Limitations: Requires Microsoft ecosystem for full functionality, occasional formatting issues with complex documents, limited standalone capabilities outside Microsoft apps
  

• Use cases in CSD: Enhancing clinical documentation efficiency, summarizing patient records, drafting professional correspondence, creating presentation materials for classes or conferences, and generating reports from assessment data
  

 

Specialized AI Tools for Academia and Healthcare

OpenKnowledgeMaps (Visual Literature Discovery)

• Features: Creates interactive visual knowledge maps of research topics, clusters related concepts visually, identifies major research areas within a field, provides direct links to open access publications, and allows exploration of conceptual relationships across disciplines
  

• Limitations: Coverage varies across different research domains, visual representation may oversimplify complex research landscapes, relies primarily on the CORE and PubMed databases, and has limited integration with reference management systems
  

• Use cases in CSD: Visualizing the landscape of research in specific communication disorders, identifying conceptual connections between treatment approaches, introducing students to research domains, planning interdisciplinary research projects, and discovering unexpected connections between communication sciences and other fields
  

Scite AI (Citation Context Analysis)

• Features: Analyzes how papers are cited by showing supporting, contrasting, or mentioning contexts, tracks citation trends over time, evaluates research impact beyond citation counts, identifies scientific consensus or controversy, and provides email alerts for new citations
  

• Limitations: More comprehensive coverage in medical sciences than in rehabilitation fields, requires institutional subscription for full functionality, citation classification accuracy varies by field, and has a learning curve for interpreting citation contexts
  

• Use cases in CSD: Evaluating the strength of evidence for clinical approaches, identifying controversial areas in communication disorders research, tracking how seminal papers in the field have been received, finding critiques of popular assessment tools, and teaching students critical research evaluation skills
  

Elicit (Research Assistant AI)

• Features: Answers research questions by analyzing scientific literature, summarizes key findings across multiple papers, extracts methodological details and sample characteristics, identifies consensus views and contradictions, and generates literature review tables automatically
  

• Limitations: Occasionally misinterprets complex research designs, has variable coverage of speech-language pathology and audiology journals, struggles with very technical terminology, and may miss nuances in qualitative research
  

• Use cases in CSD: Efficiently gathering evidence for clinical decision-making, identifying appropriate assessment measures for specific populations, comparing methodologies across intervention studies, preparing literature reviews for grant applications, and supporting evidence-based practice implementation in clinical settings
  

Research Rabbit (Literature Discovery Engine)

• Features: Creates personalized research recommendations based on saved papers, visualizes citation networks and author collaborations, tracks new publications in specific research areas, identifies seminal works and emerging trends, and integrates with reference management software
  

• Limitations: Effectiveness depends on initial paper selection quality, recommendation algorithm may create "filter bubbles" limiting exposure to divergent approaches, has less comprehensive coverage of older literature, and requires regular interaction to optimize recommendations
  

• Use cases in CSD: Discovering relevant research for thesis and dissertation projects, staying current with publications in specialized areas like swallowing disorders or cochlear implants, identifying potential research collaborators, tracking the evolution of theoretical frameworks in communication sciences, and supporting systematic literature reviews with comprehensive citation tracing
  

DALL-E/Midjourney (Image Generation)

• Features: Creates custom images from text descriptions, can generate therapy materials with specific visual requirements, allows style customization, and supports variations on existing images
  

• Limitations: Limited understanding of specialized clinical imagery, occasional anatomical inaccuracies, cannot create photorealistic images of real people, and requires careful prompt engineering
  

• Use cases in CSD: Creating custom visual supports for therapy, generating illustrated social stories, designing visual schedules and communication boards, developing patient education visuals, and creating unique materials for different cultural contexts
  

Grammarly (Writing Assistant)

• Features: Advanced grammar and spelling correction, stylistic improvement suggestions, tone adjustment capabilities, plagiarism detection, and citation assistance
  

• Limitations: Limited understanding of clinical terminology, occasional inappropriate corrections for field-specific language, and primarily focused on writing enhancement rather than content generation
  

• Use cases in CSD: Improving clarity in clinical documentation, enhancing student papers and assignments, refining research manuscripts, ensuring professional tone in communications, and standardizing departmental materials
  

 

Speech-Specific AI Tools

Speechify/Natural Reader (Text-to-Speech)

• Features: High-quality natural-sounding voice synthesis, multilingual capabilities, adjustable speaking rate and voice selection, document conversion capabilities, and mobile app integration
  

• Limitations: Limited emotional expression, occasional pronunciation errors with specialized terminology, and subscription costs for premium voices
  

• Use cases in CSD: Creating auditory stimuli for therapy, providing models for prosody and intonation, generating personalized listening exercises, supporting reading comprehension activities, and developing materials for auditory bombardment
  

Google Speech-to-Text/Microsoft Azure Speech Services (Speech Recognition)

• Features: High-accuracy transcription capabilities, speaker diarization options, custom vocabulary training, multilingual support, and integration with other applications
  

• Limitations: Variable performance with disordered speech, challenging implementation requiring technical knowledge, subscription costs scaling with usage, and privacy considerations for clinical data
  

• Use cases in CSD: Automatic transcription of assessment sessions, measuring speech metrics in therapy, creating subtitles for educational videos, documenting patient progress, and supporting self-monitoring in therapy
  

 

Emerging Specialized Tools

AI-Enhanced Diagnostic Support Tools

• Features: Pattern recognition in assessment data, comparison against normative databases, suggestion of additional assessment areas, automatic scoring assistance, and integration with electronic health records
  

• Limitations: Early development stage with limited availability, require careful validation, potential bias in training data, and need for clinician oversight
  

• Use cases in CSD: Supporting differential diagnosis, identifying subtle patterns in assessment data, reducing scoring burden, suggesting evidence-based treatment directions, and enhancing assessment efficiency
  

Virtual Patient Simulators

• Features: Interactive clinical scenarios, realistic patient responses, branching conversation paths, performance feedback mechanisms, and customizable case complexities
  

• Limitations: Limited range of simulated disorders, occasional unnatural interactions, significant development costs, and technical implementation challenges
  

• Use cases in CSD: Clinical training for students, practicing assessment techniques, developing counseling skills, preparing for challenging clinical interactions, and standardizing clinical competency evaluation
  

The landscape of AI tools available to CSD professionals continues to evolve rapidly. When selecting a platform, consider factors such as:

• Data privacy compliance: Ensure the tool meets HIPAA and educational privacy requirements
  
• Evidence base: Evaluate available research on tool effectiveness for CSD applications
  
• Accessibility: Consider interface design and compatibility with existing workflows
  
• Cost-benefit ratio: Weigh subscription costs against time savings and improved outcomes
  
• Learning curve: Factor in training time needed for effective implementation
  

For implementation success, start with a single tool addressing a specific clinical or educational need, thoroughly evaluate its performance, and gradually expand usage as expertise develops. Maintain appropriate documentation of AI usage in clinical and educational contexts according to professional guidelines.

Best Practices for Reviewing and Refining AI Outputs

Critical Evaluation Framework

• Accuracy Assessment
  
  • Cross-reference factual information with authoritative sources
    
  • Verify that assessment procedures and treatment recommendations align with current best practices
    
  • Check that terminology usage is current and consistent with ASHA guidelines
    
• Clinical Relevance Evaluation
  
  • Ensure recommendations are appropriate for the specific case details
    
  • Verify that suggested goals are functional and measurable
    
  • Confirm that cultural and individual factors are appropriately addressed
    
• Ethical Considerations
  
  • Review for potential biases in language or recommendations
    
  • Ensure confidentiality principles are maintained
    
  • Verify that limitations of AI-generated content are acknowledged when appropriate
    
• Refinement Strategy
  
  • Identify areas requiring professional judgment or nuance
    
  • Use follow-up prompts to address gaps or inaccuracies
    
  • Apply your clinical expertise to customize generic recommendations
    

Collaborative Refinement Process

Step 1: Initial Review Evaluate the AI output against these questions:

• Does this align with current evidence-based practice?
  
• Is the content clinically sound and realistic?
  
• Are there factual errors or outdated recommendations?
  
• Does the language reflect person-first, culturally responsive approaches?
  

Step 2: Targeted Refinement Use these follow-up prompt structures:

• "Revise the section on [topic] to better reflect current practice guidelines from ASHA regarding [specific area]."
  
• "The assessment recommendations need more specificity for [population]. Please suggest 3-4 standardized assessments appropriate for this case."
  
• "The language used to describe [condition/population] needs updating to reflect current terminology. Please revise while maintaining the clinical content."
  

Step 3: Final Integration

• Combine AI-generated content with your professional expertise
  
• Add specific clinical insights that AI may not capture
  
• Ensure the final product meets professional and ethical standards
  

Interactive Resources

Hands-On Demo Exercises for Faculty and Students

Exercise 1: Progressive Prompt Refinement

Learning Objective: Develop skills in iterative prompting to achieve desired outcomes

Activity Steps:

• Begin with a basic prompt: "Create a language sample analysis report"
  
• Review the output and identify three specific areas for improvement
  
• Create a refined prompt addressing those limitations
  
• Compare outputs and discuss the differences
  
• Develop a "best practice" prompt template for this task
  

Reflection Questions:

• How did specifying audience, purpose, and format affect the output?
  
• What clinical details had the greatest impact on output quality?
  
• How might you apply these prompting strategies in your clinical documentation workflow?
  

Exercise 2: Comparative Analysis Workshop

Learning Objective: Develop critical evaluation skills for AI-generated content

Materials Needed:

• AI-generated assessment report
  
• Human-written assessment report (anonymized)
  
• Evaluation rubric focusing on clinical accuracy, appropriate terminology, logical organization, and ethical considerations
  

Activity Steps:

• Review both reports without knowing which is AI-generated
  
• Score each according to the evaluation rubric
  
• Identify strengths and weaknesses of each report
  
• Reveal which report was AI-generated and discuss findings
  
• Collaborate to create guidelines for effectively using AI as a documentation assistant
  

Exercise 3: Case Study Enhancement Lab

Learning Objective: Learn to use AI as a tool for creating comprehensive teaching materials

Activity Format: Faculty and students work in pairs to:

• Begin with a basic case description (2-3 sentences)
  
• Use AI to generate expanded case details across multiple dimensions:
  
• Medical history and comorbidities
  
• Social and educational background
  
• Assessment results with specific test scores
  
• Sample communication behaviors
  
• Critically evaluate and refine the generated content
  
• Create a final case study that incorporates both AI-generated content and clinical expertise
  
• Share and critique final cases, discussing how AI assistance affected the process
  

 

AI-Generated vs. Human-Generated Content Comparisons

Comparison 1: Clinical Documentation

Sample A: Initial Evaluation Report (AI-Generated) [Provide complete sample with typical AI-generated patterns]

Sample B: Initial Evaluation Report (Human-Generated) [Provide complete sample showcasing experienced clinician documentation]

Analysis Guide:

• Structure and organization differences
  
• Depth of clinical reasoning demonstrated
  
• Use of standardized versus personalized language
  
• Integration of assessment data with recommendations
  
• Documentation of clinical decision-making
  
• Presence of nuance in prognostic statements
  

Discussion Questions:

• What subtle clinical insights appear in the human-generated report that are missing from the AI version?
  
• How might AI-generated content be modified to better capture clinical reasoning?
  
• What documentation tasks might benefit most from AI assistance?
  

Comparison 2: Patient Education Materials

Sample A: Aphasia Information Handout (AI-Generated) [Provide complete sample with typical AI-generated patterns]

Sample B: Aphasia Information Handout (Human-Generated) [Provide complete sample showcasing experienced clinician work]

Analysis Guide:

• Accessibility of language for target audience
  
• Personalization and sensitivity to patient experience
  
• Integration of practical strategies with theoretical information
  
• Visual organization and emphasis on key points
  
• Cultural sensitivity and inclusiveness
  
• Anticipation of patient questions and concerns
  

Interactive Exercise: Create a hybrid document that combines the strengths of both approaches:

• Identify 3-5 strengths from each document
  
• Draft a new document incorporating these strengths
  
• Use AI to help refine the final product
  
• Reflect on the workflow and division of labor between human expertise and AI assistance
  

Comparison 3: Lesson Plan Development

Sample A: Graduate Course Lesson Plan on Dysphagia (AI-Generated) [Provide complete sample with typical AI-generated patterns]

Sample B: Graduate Course Lesson Plan on Dysphagia (Human-Generated) [Provide complete sample showcasing experienced educator work]

Analysis Guide:

• Integration of theoretical concepts with clinical applications
  
• Anticipation of student misconceptions and learning challenges
  
• Incorporation of current research and evolving clinical practices
  
• Adaptability for different learning styles and backgrounds
  
• Assessment strategies aligned with learning objectives
  
• Opportunities for critical thinking and clinical decision-making
  

Collaborative Enhancement Activity:

• Small groups identify the strongest elements from both samples
  
• Groups use AI assistance to develop an enhanced lesson plan
  
• Faculty and students evaluate the enhanced plans using a standardized rubric
  
• Discuss how the human-AI collaboration affected the quality of the final product
  

This structured comparison approach helps CSD faculty and students develop discernment about where AI tools can enhance their work while recognizing the irreplaceable value of human clinical expertise, ethical judgment, and interpersonal connection in communication sciences and disorders practice.

Updated November 2025