{"id":31890,"date":"2025-09-25T16:27:33","date_gmt":"2025-09-25T20:27:33","guid":{"rendered":"https:\/\/www.uvm.edu\/it\/kb\/?post_type=ht_kb&p=31890"},"modified":"2025-11-05T15:33:07","modified_gmt":"2025-11-05T20:33:07","slug":"best-practices-for-copilot-llm-prompting","status":"publish","type":"ht_kb","link":"https:\/\/www.uvm.edu\/it\/kb\/article\/best-practices-for-copilot-llm-prompting\/","title":{"rendered":"Best Practices for Copilot\/LLM Prompting"},"content":{"rendered":"

Why It Matters<\/span><\/b>\u00a0<\/span><\/p>\n

The quality of your AI output depends heavily on how you craft your prompts. Think of prompting as \u201cprogramming with words\u201d \u2014 clear, structured input leads to better results and a reduced number of calls.<\/span>\u00a0<\/span><\/p>\n \t\t

\r\n \t\t\t

UVM-Specific Tips<\/h3><\/div>\r\n \t\t\t
\r\n \t\t\t\t
\r\n \t\t\t\t\t<\/p>\n
    \n
  • Respect Data Policies<\/span><\/b>: Use your UVM account (look for the <\/span>Green Shield Icon<\/span><\/i> to ensure Enterprise Data Protection is enabled)<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
      \n
    • Verify Outputs<\/span><\/b>: Always fact-check AI responses before sharing.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
        \n
      • Understand Limitations<\/span><\/b>: LLMs may hallucinate; use them as assistants, not authorities.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
          \n
        • Prompt Gallery<\/span><\/b>: Save and reuse effective prompts via Copilot\u2019s Prompt Gallery feature for consistency.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n

          \n \t\t\t\t<\/div>\r\n \t\t\t<\/div>\r\n \t\t<\/div>\r\n \t\t\n \t\t

          \r\n \t\t\t

          Core Principles<\/h3><\/div>\r\n \t\t\t
          \r\n \t\t\t\t
          \r\n \t\t\t\t\t<\/p>\n

          Provide Context<\/span><\/b>\u00a0<\/span><\/p>\n

            \n
          • Explain the background, audience, and purpose.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
              \n
            • Example:\u202f<\/span>\u201cWrite a 200-word summary of this report for a non-technical audience.\u201d<\/span><\/i>\u00a0<\/span><\/li>\n<\/ul>\n

              Be Specific<\/span><\/b>\u00a0<\/span><\/p>\n

                \n
              • Include details like tone, format, and constraints.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                  \n
                • Example:\u202f<\/span>\u201cGenerate five social media captions under 100 characters, using a friendly tone.\u201d<\/span><\/i>\u00a0<\/span><\/li>\n<\/ul>\n

                  Use Step-by-Step Instructions<\/span><\/b>\u00a0<\/span><\/p>\n

                    \n
                  • Break complex tasks into smaller steps.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                      \n
                    • Example:\u202f<\/span>\u201cFirst outline the key points, then draft a summary.\u201d<\/span><\/i>\u00a0<\/span><\/li>\n<\/ul>\n

                      Iterate and Refine<\/span><\/b>\u00a0<\/span><\/p>\n

                        \n
                      • Start broad, then refine based on the AI\u2019s response.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                          \n
                        • Use follow-up prompts to clarify or adjust.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n

                          Leverage Examples<\/span><\/b>\u00a0<\/span><\/p>\n

                            \n
                          • Provide sample outputs or templates.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                              \n
                            • Example:\u202f<\/span>\u201cHere\u2019s an example of the tone I want: [insert example].\u201d<\/span><\/i>\u00a0<\/span><\/li>\n<\/ul>\n

                              Set Role or Perspective<\/span><\/b>\u00a0<\/span><\/p>\n

                                \n
                              • Assign the AI a role for better alignment.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                                  \n
                                • Example:\u202f<\/span>\u201cAct as a cybersecurity analyst and explain the risks of weak passwords.\u201d<\/span><\/i>\u00a0<\/span><\/li>\n<\/ul>\n

                                  \n \t\t\t\t<\/div>\r\n \t\t\t<\/div>\r\n \t\t<\/div>\r\n \t\t\n \t\t

                                  \r\n \t\t\t

                                  Common Pitfalls<\/h3><\/div>\r\n \t\t\t
                                  \r\n \t\t\t\t
                                  \r\n \t\t\t\t\t<\/p>\n
                                    \n
                                  • Vague prompts like\u202f<\/span>\u201cWrite something about AI\u201d<\/span><\/i>\u202f\u2192 lead to generic results.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                                      \n
                                    • Overloading with multiple unrelated tasks in one prompt.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n
                                        \n
                                      • Ignoring iterative refinement.<\/span>\u00a0<\/span><\/li>\n<\/ul>\n

                                        \n \t\t\t\t<\/div>\r\n \t\t\t<\/div>\r\n \t\t<\/div>\r\n \t\t\n","protected":false},"excerpt":{"rendered":"

                                        Why It Matters\u00a0 The quality of your AI output depends heavily on how you craft your prompts. Think of prompting as \u201cprogramming with words\u201d \u2014 clear, structured input leads to better results and a reduced number of calls.\u00a0<\/p>\n","protected":false},"author":111,"comment_status":"closed","ping_status":"closed","template":"","format":"standard","meta":{"_relevanssi_hide_post":"","_relevanssi_hide_content":"","_relevanssi_pin_for_all":"","_relevanssi_pin_keywords":"","_relevanssi_unpin_keywords":"","_relevanssi_related_keywords":"","_relevanssi_related_include_ids":"","_relevanssi_related_exclude_ids":"","_relevanssi_related_no_append":"","_relevanssi_related_not_related":"","_relevanssi_related_posts":"","_relevanssi_noindex_reason":"","_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"footnotes":""},"ht-kb-category":[272],"ht-kb-tag":[625,519],"class_list":["post-31890","ht_kb","type-ht_kb","status-publish","format-standard","hentry","ht_kb_category-software-and-licensing","ht_kb_tag-copilot","ht_kb_tag-microsoft"],"_links":{"self":[{"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb\/31890","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb"}],"about":[{"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/types\/ht_kb"}],"author":[{"embeddable":true,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/users\/111"}],"replies":[{"embeddable":true,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/comments?post=31890"}],"version-history":[{"count":4,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb\/31890\/revisions"}],"predecessor-version":[{"id":31908,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb\/31890\/revisions\/31908"}],"wp:attachment":[{"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/media?parent=31890"}],"wp:term":[{"taxonomy":"ht_kb_category","embeddable":true,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb-category?post=31890"},{"taxonomy":"ht_kb_tag","embeddable":true,"href":"https:\/\/www.uvm.edu\/it\/kb\/wp-json\/wp\/v2\/ht-kb-tag?post=31890"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}