Lesson 7 of 7
AI Challenge
Synthesise your knowledge across machine learning, generative models, and ethics to solve real-world AI problems.
Learn it
You have made it to the final challenge! You now understand what AI is, how models learn from data, how generative tools paint images and write text, and why ethics matter.
In this ultimate mission, you will step into the shoes of an AI Systems Architect. You will assemble all these puzzle pieces to design smart, safe, and fair AI solutions for the future.
Key terms
- AI Architect
- An engineer who designs end-to-end artificial intelligence systems, data pipelines, and safety protocols.
- Retrieval-Augmented Generation (RAG)
- A technique that connects an LLM to verified external databases to provide accurate, grounded responses.
- Fine-Tuning
- The process of taking a pre-trained model and further training it on a specialised domain dataset.
- Alignment
- Guiding AI systems so that their decisions, outputs, and behaviours remain safe and aligned with human values.
Designing an End-to-End AI Solution
Walk through the complete architectural blueprint for launching a medical triage AI assistant safely.
- 1Problem Definition and Data Scoping: Define clinical boundaries and collect anonymised, balanced patient medical records with strict data privacy consent.
- 2Model Selection and RAG Integration: Select a pre-trained foundation model and connect it via RAG to verified, peer-reviewed medical databases.
- 3Safety Guardrails and Moderation: Build safety filters that instantly flag emergency conditions and route urgent cases directly to human emergency doctors.
- 4Bias Audit and Red-Teaming: Employ diverse clinical teams to aggressively test the system for diagnostic disparities across patient demographics.
- 5Continuous Human-in-the-Loop Deployment: Deploy the assistant as a decision-support tool where medical professionals verify every automated recommendation.
Guardrail Implementation in Python
python# Medical triage safety guardrail check
def process_query(user_query, urgent_keywords):
for keyword in urgent_keywords:
if keyword in user_query.lower():
return 'ALERT: Critical symptom detected. Transferring to emergency doctor.'
return 'Query routed to medical retrieval database.'
emergency_terms = ['chest pain', 'cannot breathe', 'severe bleeding']
patient_input = 'I am experiencing severe chest pain after running.'
print(process_query(patient_input, emergency_terms))Production AI architectures implement safety filters and guardrails before queries reach generative models, intercepting critical risks immediately.
Try it
Order the steps required to safely build and deploy a specialised customer service AI assistant.
- 1Implement a RAG pipeline connecting the LLM to official documentation
- 2Configure safety filters and red-team the system for harmful outputs
- 3Launch with human-in-the-loop oversight and active telemetry monitoring
- 4Conduct demographic bias testing and performance benchmarks
- 5Define system boundaries and collect verified domain knowledge documents
Challenge
Design an AI system for a marine conservation charity to protect endangered coral reefs. Detail the sensor inputs, model types (e.g. computer vision vs time-series prediction), and ethical safeguards against false alarms.
Pick whichever way suits you — every mode earns the same bonus XP.
Write at least 40 more characters to submit.
Mark your own work
Guided walkthrough — 0/5 clues revealed
- Clue 1 locked — reveal it only if you get stuck.
- Clue 2 locked — reveal it only if you get stuck.
- Clue 3 locked — reveal it only if you get stuck.
- Clue 4 locked — reveal it only if you get stuck.
- Clue 5 locked — reveal it only if you get stuck.
Each clue costs 6 XP (never below 28 XP). You'd earn 55 XP right now.
Extension: How would you ensure local fishing communities can inspect and trust the system recommendations?