Understanding AI Responses

PioneerChat responses include several types of content. Understanding these helps you get the most from the AI and verify the information you receive.

Response Components

Main Response

The primary answer to your question, formatted with:

• Headings and subheadings for organisation
• Bullet points and numbered lists
• Bold and italic text for emphasis
• Code blocks for technical content

Reasoning Blocks

When enabled (or for complex questions), you may see a "Thinking" or "Reasoning" section. This shows the AI's thought process before giving its final answer.

Reasoning blocks are collapsible - click to expand and see the full thought process. This is useful for:

• Understanding how the AI reached its conclusion
• Verifying the logic behind recommendations
• Learning about the AI's approach to problems

Tool Blocks

When the AI uses tools (like document analysis or calculations), you'll see tool blocks showing what it's doing. These indicate:

• Tool name: Which capability is being used
• Parameters: What inputs were provided
• Status: Whether the tool is running or complete

Streaming Responses

PioneerChat streams responses in real-time, meaning you see the answer as it's being generated. You'll notice:

• Text appearing progressively
• A "Generating..." indicator while processing
• Thinking blocks may show a pulsing animation while active

You can start reading immediately - you don't need to wait for the full response.

Providing Feedback

Each AI response has feedback buttons (thumbs up/down). Use these to:

• Thumbs up: The response was helpful and accurate
• Thumbs down: The response was unhelpful, inaccurate, or problematic

Your feedback helps improve the service. When you give negative feedback, consider following up with what was wrong so the AI can try again.

Limitations to Understand

The AI May:

• Not have current information: Knowledge has a cutoff date
• Make confident-sounding errors: Verify important facts
• Misinterpret ambiguous requests: Be specific in your prompts
• Generate plausible but incorrect details: Especially for specific dates, statistics, or quotes

Best Practices:

• Double-check facts for important documents
• Review calculations independently when accuracy is critical
• Ask the AI to cite sources when available
• Use reasoning mode to see the AI's logic

When the AI Says "I Don't Know"

The AI may acknowledge uncertainty. This is actually a good sign - it's being honest rather than making something up. When this happens:

• Try rephrasing your question
• Provide more context or documents
• Break the question into smaller parts
• Check if the information might be in an uploaded document

Understanding "I Can't Help With That"

The AI has guidelines about what it can assist with. It may decline requests that:

• Could cause harm
• Involve illegal activities
• Request personal information about individuals
• Fall outside its capabilities

If you believe the AI has misunderstood your request, try rephrasing to clarify your legitimate intent.