Logo for AiToolGo

Getting Started with AI: Use Cases and Prompt Tips for Harvard Users

Overview
Easy to understand
 0
 0
 1
Harvard University Information Technology (HUIT) guide introducing basic AI tool use cases—content creation and review, search (including deep research features), data analysis, and accessibility—each paired with cautions about inaccuracy and the need for human verification. It also offers prompt-writing guidance: be specific about goals, constraints, context, tone, and output format; break requests into iterative chunks; give feedback and corrections; and ask the AI for help clarifying what it needs. Includes links to HUIT support resources and reminders to use approved tools per University guidelines.
  • main points
  • unique insights
  • practical applications
  • key topics
  • key insights
  • learning outcomes

  • • main points

    • 1
      Clear, well-organized coverage of four core AI use cases with consistent 'Keep in mind' risk warnings for each
    • 2
      Practical, actionable prompting framework (goal, boundaries, context, tone/audience, output format) that beginners can apply immediately
    • 3
      Emphasizes iterative prompting and human verification, promoting responsible and accurate AI use in an institutional setting

  • • unique insights

    • 1
      Frames prompting as a conversation with a colleague: iterate, give feedback, and ask the AI what information it needs to perform well
    • 2
      Highlights institution-specific considerations such as using only approved tools, data privacy, and accessibility services like Harvard's Digital Accessibility Services

  • • practical applications

    • Serves as a concise onboarding primer for non-expert university staff and students, giving immediate, usable prompting techniques and risk-awareness habits, though it offers limited depth on advanced features and no hands-on tool-specific walkthroughs.

  • • key topics

    • 1
      Basic AI use cases: content creation, search, data analysis, accessibility
    • 2
      Prompt engineering fundamentals: specificity, context, tone, format, iteration
    • 3
      Responsible AI use: verifying outputs, institutional guidelines, approved tools

  • • key insights

    • 1
      Concise, institution-grounded introduction that pairs each capability with its specific failure modes
    • 2
      A ready-to-use prompt structure tailored for everyday professional tasks rather than technical developers

  • • learning outcomes

    • 1
      Identify four core everyday use cases for AI tools and their associated limitations
    • 2
      Write clearer, more specific prompts that include goals, constraints, context, audience, and output format
    • 3
      Apply an iterative, feedback-driven approach to refining AI outputs and verifying results before use
examples
tutorials
code samples
visuals
fundamentals
advanced content
practical tips
best practices

“ Introduction: Getting Started with AI at Harvard

Artificial intelligence tools are becoming part of everyday academic and administrative work. Before using any AI tool in the Harvard community, it is essential to review the University's guidelines and choose an approved tool. This guide offers a high-level overview of common AI use cases and practical advice for writing effective prompts, so that you can get useful results while maintaining accuracy, integrity, and accountability.

“ Content Creation and Review

AI can generate ideas and drafts for text, documents, images, presentations, music, videos, and surveys within seconds, based on simple natural language instructions. It can also act as an editing partner, helping you refine your writing, spot errors, and identify areas for improvement. Even when output looks polished and professional, it may be inaccurate, generic, biased, or include copyrighted material. University guidelines require that all AI-generated content be reviewed before it is published or shared.

“ AI-Powered Search and Deep Research

AI tools can search thousands of sources across the web quickly. Because they accept natural language, you can describe what you are looking for even when you do not know the precise terminology. Some tools offer a deep research feature, in which an AI agent works in the background to produce a detailed report on a chosen topic, citing sources along the way. Keep in mind that results may be incorrect, outdated, or irrelevant. A citation does not guarantee that the AI understood or summarized the source accurately, so verify important information against the original materials.

“ Data Analysis and Document Summarization

AI can condense long documents into concise summaries, making it easier to highlight key insights from research papers, reports, emails, meeting notes, or survey and spreadsheet data. Some tools include source-grounded or notebook features that let you upload a defined set of files or links and then analyze them, summarize them, or generate content such as presentations based on those sources. However, AI may overlook important context, misread data, make calculation errors, or draw conclusions the source does not support. Double-check key findings and calculations against the original material.

“ Improving Accessibility with AI

AI can help make content more accessible by detecting problems in documents, suggesting alt text and audio descriptions for images, and converting speech to text for captions. AI-powered accessibility checkers save time, but they cannot catch every issue. Manual testing remains necessary. Harvard's Digital Accessibility Services can help you navigate both the tools and the requirements.

“ Why Verification Matters: Key Limitations to Keep in Mind

Across all use cases, the same principle applies: AI output is a starting point, not a final answer. Errors can appear in facts, citations, calculations, interpretations, and tone. Treat AI as a capable assistant whose work you must review, check against trustworthy sources, and take responsibility for before sharing it.

“ How AI Prompts Work

AI tools analyze your inputs, known as prompts, and generate responses based on patterns learned during training. Modern AI tools understand conversational language far better than earlier systems did, so the goal is less about crafting a perfect prompt and more about clearly communicating your objective and how the AI can help you reach it.

“ Writing Specific Prompts

Generic prompts tend to produce generic results. To get better output, state clearly what you want, such as the goal or task (for example, producing a report, analyzing data, or finding information on a topic). Include what you do not want and any boundaries that keep the output focused. Provide relevant background context that helps the AI understand the purpose of the request. Specify the tone, audience, and style, such as a marketing brief, a lesson plan, or a financial report. Finally, indicate the desired output format, such as a text document, spreadsheet, presentation, or poster.

“ Refining Results Through Conversation and Feedback

You do not need to include everything in your first prompt. Treat the interaction as an ongoing conversation. Begin with a basic request and build on it, adjusting wording, tone, or detail as needed. You can upload excerpts rather than entire documents to keep the discussion focused. Give feedback on what works and what does not, and point out errors so the AI can correct them, much as you would with a colleague on a shared project.

“ Getting Help and Next Steps

If you are unsure how to write an effective prompt, ask the AI for help. Describe your goal, even if it is still taking shape, and ask what approach would work best or what information it needs from you to perform the task well. For technical support, contact the HUIT Service Desk at 617-495-7777 or ithelp@harvard.edu. You can also share feedback on AI tools and request new features through the HUIT channels provided on the site.

 Original link: https://www.huit.harvard.edu/ai-basics

Comment(0)

user's avatar

      Similar Learning

      Related Tools