You don't trust AI-generated information - strategies to verify and cross-check AI outputs
When learning with AI tools like Wise-Owl, it's natural and appropriate to question the reliability of the information you're receiving. Critical thinking includes being sceptical of your sources, and AI should be no exception. Not all information has the same level of risk if it's incorrect—some facts are high-stakes and require thorough verification, while others are low-stakes and safe to accept provisionally. The strategies below will help you develop a systematic approach to verifying AI-generated information, distinguish between high-stakes and low-stakes information, and build confidence in using AI as a learning tool while maintaining appropriate scepticism.
Each strategy starts with a flow diagram showing the order in which modes and functions are used, followed by an explanation of how this strategy would work for your issue and then examples of prompts or question you can type or paste into the chat window.
Have a look at these four and choose one that suits your situation best. Remember, Wise-Owl is here to help you with your study; it does not do your study for you. Learning is always effortful, even with AI, but it is an investment in your future. Your degree might help you getting into the labour market, but the quality of your learning is what makes you successful in your career.
Strategy A
Strategy A: Systematic Verification Protocol
· Flow: Tutor Mode → Conversation Starters → web search (external verification) → Tutor Mode → Deep Dive (cross-reference) → Coach Mode → Study Notes
· How it works: Learn from AI, identify key claims that need verification, use external sources to verify, then consolidate verified information
Example prompts:
o In Tutor Mode: "Explain [topic] to me"
o "What are the key facts or principles I should verify from external sources?"
o Use web search or library resources to verify the most important claims
o Return to Deep Dive: "I've verified [these facts]. Now help me understand the connections more deeply"
o In Study Notes: "Help me create notes that distinguish between AI-provided information and externally verified facts"
Strategy B
Strategy B: Risk assessment and selective verification
· Flow: Tutor Mode → Conversation Starters → Assessor Mode → Critical Reasoning → Coach Mode → Study Planner
· How it works: Learn to categorise information by stakes, verify high-stakes information, accept low-stakes information provisionally
Example prompts:
o In Tutor Mode: "Teach me about [topic]"
o "Which aspects of this information are high-stakes (critical for exams, safety-critical, or career-critical)?"
o "Which aspects are low stakes (background information, general context)?"
o Verify high-stakes information using authoritative external sources (textbooks, academic papers, official documentation)
o In Study Planner: "Create a verification checklist for high-stakes information in [topic]"
Strategy C
Strategy C: Comprehensive trust-building system
· Flow: Tutor Mode → Conversation Starters (with verification prompts) → Coach Mode → Study Notes (flagged items) → web search → Tutor Mode → Deep Dive (verification discussion) → Assessor Mode → Applied Knowledge → Simulator Mode → Role Play → Coach Mode → Portfolio Coach → Tutor Mode → Summarise
· How it works: Learn with verification mindset, flag items for checking, verify externally, discuss verification, test understanding, apply in scenarios, reflect on verification process, consolidate verified knowledge
Example prompts:
o Tutor Mode → Conversation Starters: "Explain [topic] and highlight which claims I should verify independently"
o Coach Mode → Study Notes: "Help me create notes with a 'verify' flag for high-stakes claims"
o Use external resources to verify flagged claims
o Tutor Mode → Deep Dive: "I verified [claims]. How do these verified facts fit into the bigger picture?"
o Assessor Mode → Applied Knowledge: "Test me on the verified information"
o Simulator Mode → Role Play: "Create a scenario where incorrect information would have serious consequences"
o Coach Mode → Portfolio Coach: "Reflect on my verification process. How can I improve?"
o Tutor Mode → Summarise: "Summarise the verified understanding of [topic]"
Strategy D
Strategy D: Multi-source triangulation
· Flow: Tutor Mode → Magic Question → Coach Mode → Study Notes (source tracking) → web search → Simulator Mode → Study Group (compare sources) → Assessor Mode → Critical Reasoning → Tutor Mode → Deep Dive → Coach Mode → Portfolio Coach
· How it works: Get AI explanation, track what needs verification, find multiple external sources, compare all sources including AI, think critically about discrepancies, deepen understanding with awareness of source reliability, reflect on evaluation process
Example prompts:
o Tutor Mode → Magic Question: "What are the most important claims about [topic] that I should verify?"
o Coach Mode → Study Notes: "Help me track claims and their sources (AI + external sources)"
o Find 2-3 authoritative sources on the topic
o Simulator Mode → Study Group: "I have information from [AI, Source 1, Source 2]. Let's compare and identify agreements and discrepancies"
o Assessor Mode → Critical Reasoning: "Analyse why these sources might differ on [specific point]"
o Tutor Mode → Deep Dive: "Given the verified facts, explain [topic] with appropriate caveats"
o Coach Mode → Portfolio Coach: "Reflect on my source evaluation skills"