Create Rule “avoid-ai-hallucinations/rule” (#10018)

* Create Rule “avoid-ai-hallucinations/rule”

* Update Rule “avoid-ai-hallucinations/rule”

* Update Rule “avoid-ai-hallucinations/rule”

* Update Rule “avoid-ai-hallucinations/rule”

* Update Rule “avoid-ai-hallucinations/rule”

* Add to rule

* Update rule.md

---------

Co-authored-by: Tiago Araújo [SSW] <[email protected]>

Author: edkranz

categories/artificial-intelligence/rules-to-better-ai-development.md

@@ -18,6 +18,7 @@ index:
 - use-embeddings
 - best-ai-powered-ide
 - ai-for-frontend-development
+- avoid-ai-hallucinations
 
 ---
 

rules/avoid-ai-hallucinations/rule.md

@@ -0,0 +1,131 @@
+---
+type: rule
+tips: ""
+title: Do you handle AI Hallucinations the right way?
+seoDescription: AI hallucinations are inevitable, but with the right techniques, you can minimize their occurrence and impact.
+uri: avoid-ai-hallucinations
+authors:
+  - title: Eddie Kranz
+    url: https://www.ssw.com.au/people/eddie-kranz/
+related:
+  - rules-to-better-chatgpt-prompt-engineering
+created: 2025-03-17T14:57:00.000Z
+guid: e4e963e4-1568-4e47-b184-d2e96bc0f124
+---
+AI is a powerful tool, however, sometimes it simply makes things up, aka hallucinates. While hallucinating in your spare time is pretty cool, it is very bad for business!
+
+AI hallucinations are inevitable, but with the right techniques, you can minimize their occurrence and impact. Learn how SSW tackles this challenge using proven methods like clean data tagging, multi-step prompting, and validation workflows.  
+
+<!--endintro-->
+
+**Let's face it. AI will always hallucinate.**
+
+AI models like GPT-4 are powerful but imperfect. They generate plausible-sounding but incorrect or nonsensical outputs (hallucinations) due to training limitations, ambiguous prompts, or flawed data retrieval. While you can’t eliminate hallucinations entirely, you can **reduce their frequency** and **mitigate risks**.  
+
+- - -
+
+## Use Clean, Tagged Data for RAG
+
+❌ Bad: Untagged data in a RAG system 
+
+```phyton
+documents = ["Sales grew 10% in 2023", "Server downtime: 5hrs in Q2"]  
+```
+
+::: greybox
+**Query:** "What was the server uptime in Q2?"  
+**Hallucination:** "Server uptime was 95%." (Wrong: No uptime data exists!)  
+:::
+::: bad
+Figure: Bad example - Untagged, messy data leads to garbage outputs
+:::
+
+✅ Good: Properly tagged data  
+
+```phyton
+documents = [  
+  {"text": "Sales grew 10% in 2023", "tags": ["finance", "sales"]},  
+  {"text": "Server downtime: 5hrs in Q2", "tags": ["IT", "downtime"]}  
+]
+```
+
+::: greybox  
+# Query: "What was the server uptime in Q2?"  
+# Output: "No uptime data found. Available data: 5hrs downtime." ✅  
+```
+::: good
+Figure: Good example - Properly tagged data reduces the risk of incorrect retrieval
+:::
+
+## Break Workflows into Multi-Step Prompts
+
+Use a **chain-of-thought** approach to split tasks into smaller, validated steps
+
+::: greybox
+**User:** "Write a blog about quantum computing benefits for SMEs."  
+**AI:** (Hallucinates fictional case studies and stats) 
+:::
+::: bad
+Figure: Bad example - A single-step prompt invites hallucinations
+:::
+
+::: greybox
+**User:** "Generate a blog draft about quantum computing for SMEs."\
+"Verify all claims in this draft against trusted sources."\
+"Compare the final draft to the original query. Did you answer the question?"  
+:::
+::: good
+Figure: Good example - Multi-step validation reduces errors
+::: 
+
+## Force the AI to Justify Its Reasoning
+
+Always prompt the AI to **cite sources** and **flag uncertainty**. 
+
+::: greybox
+**User:** "Why should SMEs adopt quantum computing?"  
+**AI:** "It boosts efficiency by 200%." (Source? None!)  
+:::
+::: bad
+Figure: Bad example - No justification = unchecked errors
+:::
+
+::: greybox
+**System Prompt:** "Answer the question and cite sources. If uncertain, say 'I don’t know'."  
+**User:** "Why should SMEs adopt quantum computing?"  
+**AI:** "Quantum computing can optimize logistics (Source: IBM, 2023). However, adoption costs may be prohibitive for SMEs."  
+:::
+::: good
+Figure: Good example - Require citations and self-reflection
+:::
+
+## Validate Outputs Against the Original Question
+Use a **validation layer** to ensure outputs align with the original query.  
+
+::: greybox
+**User:** "How does Azure Kubernetes Service (AKS) simplify deployment?"  
+**AI:** Explains Kubernetes basics (ignores AKS specifics). 
+:::
+::: bad
+Figure: Bad example - No final check = off-topic answers
+:::
+
+::: greybox
+System Prompt: "Compare your answer to the user’s question. Did you address AKS?"  
+AI: "Revised answer: AKS simplifies deployment by integrating with Azure DevOps and..." ✅  
+:::
+::: good
+Figure: Good example - Add a final validation step
+:::
+
+### **Other techniques to minimize hallucinations
+
+* **Lower temperature settings**: Reduce creativity (e.g., `temperature=0.3`) for factual tasks
+* **Human-in-the-loop**: Flag low-confidence responses for manual review
+* **Predefined constraints**: Example: "Do not speculate beyond the provided data"
+
+---
+
+AI hallucinations are unavoidable, but SSW’s proven techniques, like clean data tagging, multi-step validation, and forcing justification can keep them in check. By designing workflows that **anticipate errors** and **validate outputs**, you turn a risky limitation into a manageable challenge.  
+
+Always assume hallucinations **will** happen, so build systems to catch them!