Artificial Intelligence Starter · free course

Free course · Lesson 5 of 5

AI Ethics, Bias & Responsible Use

In this lesson you will

  • Identify risks of bias and privacy loss
  • Apply simple rules for responsible AI use

AI systems are only as good as the data they are trained on. If the data contains bias — for example, mostly one gender or one ethnic group — the AI will learn and amplify that bias. AI Ethics is about designing and using AI in ways that are fair, transparent, and accountable.

  • Bias in AI: A facial recognition system trained mostly on light-skinned faces may fail to recognize dark-skinned faces. This is a real problem that has been documented.
  • Data privacy: AI models should not store or share personal information without consent.
  • Transparency: Users should know when they are interacting with an AI, not a human.
  • Accountability: Someone must be responsible for what an AI system does — you cannot blame the algorithm.

💡 Practical Skill: When using AI tools, always ask: "Who trained this model? What data was used? Could this output be biased?" Test AI systems with diverse inputs to check for fairness. Use tools like IBM AI Fairness 360 or Google's What-If Tool to audit models.

Worked exampleDetecting Bias in a Dataset

Python

Python
# Simulating a bias check on a hiring dataset# Check if the dataset is balanced across genders def check_dataset_bias(data):    total = len(data)    counts = {}        for item in data:        category = item.get('gender', 'unknown')        counts[category] = counts.get(category, 0) + 1        print("Dataset Distribution:")    for category, count in counts.items():        percentage = (count / total) * 100        print(f"  {category}: {count} ({percentage:.1f}%)")        # Check if any group is underrepresented (less than 30%)    for category, count in counts.items():        percentage = (count / total) * 100        if percentage < 30 and percentage > 0:            print(f"\n⚠️ Warning: {category} is only {percentage:.1f}% of the dataset.")            print("  The AI model may perform poorly for this group!")        print("\n✅ Bias check complete.") # Example hiring datasethiring_data = [    {"name": "Applicant A", "gender": "male", "hired": True},    {"name": "Applicant B", "gender": "male", "hired": True},    {"name": "Applicant C", "gender": "male", "hired": False},    {"name": "Applicant D", "gender": "female", "hired": True},    {"name": "Applicant E", "gender": "female", "hired": False},    {"name": "Applicant F", "gender": "male", "hired": True},    {"name": "Applicant G", "gender": "male", "hired": True},    {"name": "Applicant H", "gender": "male", "hired": False},    {"name": "Applicant I", "gender": "male", "hired": True},    {"name": "Applicant J", "gender": "female", "hired": False},] check_dataset_bias(hiring_data)# Output will show gender distribution and flag any imbalance

Practice exercise

Write three rules for your workplace about what information must never be typed into public AI tools, and why.

Key takeaways

  • AI can repeat unfairness found in its data
  • Never enter personal or confidential data into unapproved tools
  • You remain responsible for AI-assisted work
In the full program Module 5: AI Ethics and Responsible Use See how this topic is taught in depth, with practical activities and an assessment.

Free self-study lesson — no certificate. For the full instructor-led program, see registration options.