​

Reporting Inaccurate AI Responses

The AI Research Assistant is powerful, but it’s not perfect. When you encounter an inaccurate response, reporting it helps us improve the AI for everyone. This guide explains how to report issues effectively and what happens after you do.

​

Why Reporting Matters

Your feedback directly impacts AI quality. When you report an issue:

• Our team reviews it to understand what went wrong
• We identify patterns across multiple reports to find systemic issues
• We improve the AI through training updates and guardrails
• Data sources get verified if the issue stems from incorrect underlying data

Every report makes the AI a little smarter and more reliable for the entire Rallies community.

​

Types of Issues Worth Reporting

​

Factually Incorrect Information

The AI stated something demonstrably false, such as wrong financial figures, incorrect company information, or misattributed quotes. Example: “The AI said Apple’s P/E ratio is 45, but the stock page shows it’s 28.”

​

Outdated Information

The AI provided old data when newer data exists, like referencing last quarter’s earnings when current quarter is available. Example: “The AI said the CEO is John Smith, but he stepped down last month.”

​

Hallucinations

The AI made up information that doesn’t exist, such as invented statistics, fake analyst quotes, or fabricated events. Example: “The AI claimed there was a ‘Goldman Sachs report from March’ rating this stock, but I can’t find any such report anywhere.”

​

Calculation Errors

The AI made mathematical mistakes, like incorrect percentage calculations or wrong ratios. Example: “The AI said revenue grew 50% from $8B to$10B, but that’s actually only 25% growth.”

​

Misinterpretation

The AI understood your question wrong, answered a different question, or confused similar companies.

​

How to Report an Issue

​

Method 1: Feedback Buttons (Fastest)

Every AI response has feedback buttons at the bottom:

1. Find the thumbs up/down icons below the AI’s response
2. Click the thumbs down if the response was problematic
3. A feedback form appears - fill in what went wrong
4. Submit your feedback

This links your feedback directly to the specific response.

​

Method 2: In-Conversation Reporting

Tell the AI directly about the error in your next message. The AI will acknowledge and try to correct. Then use the feedback button on the corrected response.

​

Method 3: Contact Support

For serious or recurring issues, email [email protected] with the subject “AI Accuracy Issue” and include:

• The question you asked
• The AI’s response (copy/paste or screenshot)
• What was wrong and the correct information
• Source for the correct information (if you have one)

Contact support →

​

What to Include in Your Report

​

Example of a Good Report

Question I asked: “What is Tesla’s debt-to-equity ratio?” AI Response: “Tesla’s debt-to-equity ratio is 1.8” What’s wrong: The ratio is incorrect. According to Tesla’s latest 10-Q and the Rallies stock page, the debt-to-equity ratio is 0.11.

​

What Happens After You Report

When you use the feedback button, your report is logged immediately. Our team reviews reports regularly through this process:

1. Triage - We categorize the issue (data error, AI error, user misunderstanding)
2. Investigation - We reproduce the issue and identify the root cause
3. Verification - We confirm the correct information
4. Resolution - We implement fixes (data correction, model update, guardrail addition)

Due to volume, we typically can’t respond individually to feedback reports. However, every report is reviewed and patterns across reports prioritize fixes.

​

Best Practices for Accurate Information

​

Cross-Reference with Stock Pages

The stock pages pull directly from our data sources. Check the numbers the AI provides against the stock page for verification.

​

Verify Important Figures

Before making decisions based on AI-provided data, check SEC filings (EDGAR database), company investor relations pages, or reputable financial news.

​

Ask for Sources

You can ask the AI where it got its information:

“What’s your source for that revenue figure?”

​

Be Skeptical of Specifics

Be extra careful with very specific numbers, quotes attributed to people, predictions, or information about very recent events.

​

Common Causes of AI Errors

Data Timing: Financial data updates quarterly. There’s a lag between events and data updates. AI Limitations: AI can hallucinate plausible-sounding but false information. Complex calculations may have errors. Ambiguity: Your question may have multiple interpretations, or company names can be confused. Data Source Variations: Different sources may report slightly different figures (adjusted vs. unadjusted, GAAP vs. non-GAAP). Learn more about AI limitations →

---

​

Frequently Asked Questions

​

Does reporting actually change anything?

Yes. User reports directly influence our improvement priorities. Patterns in reports help us identify systematic issues, and detailed reports often lead to specific fixes.

​

What if I’m not sure if it’s wrong?

Report it anyway. Include what you think might be wrong and why you’re uncertain. This helps us understand where users have doubts.

​

What if the AI keeps making the same mistake?

If you notice a recurring error, email [email protected] with the pattern you’ve observed. Systemic issues get higher priority than one-off errors.

---

​

Related Articles

• Understanding AI Limitations
• How the AI Research Assistant Works
• Getting Better Answers: Prompt Tips
• How to Contact Support