AI Ethics in Your Hands

Principles & Practice

Protect Yourself, Respect Others, Shape the Future

Fact: 72% of internet users worry about AI misuse but only 31% take steps to protect themselves

Source: Pew Research, 2024

Worried: 72% Taking Action: 31%

Before We Begin...

If you know Mathematics, Pure → Applied.

You can define AI with Ethics.

Otherwise, through life read AI and its Ethics.

Mathematical Foundation

AI Ethics?

Definition:

AI Ethics is the set of principles, guidelines, and practices that ensure Artificial Intelligence is developed and used in ways that are fair, transparent, safe, and beneficial to individuals and society.

Key Focus Areas:

Fairness

Avoid bias and discrimination

Transparency

Make AI decisions explainable

Privacy

Protect user data and consent

Accountability

Assign responsibility for AI outcomes

Safety

Prevent harm from AI misuse

Example:

When an AI model is used for hiring, AI Ethics ensures it doesn't unfairly reject candidates based on gender, ethnicity, or disability — and that the hiring process can be reviewed by humans.

Why AI Ethics?

Economic Impact

AI could add $15.7 trillion to the global economy by 2030

Source: PwC, 2023

Risks Without Ethics

  • Privacy breaches
  • Bias and discrimination
  • Misinformation spread
  • Physical and psychological harm

Unethical AI can lead to wrongful decisions, loss of trust, and legal consequences for organizations.

Energy Consumption: ChatGPT vs Everyday Activities

Energy Consumption Comparison

Activity Energy (Wh) Equivalent Time
ChatGPT Query (Simple) 0.3 Wh 0.3 min of 60W bulb
LED Bulb (10W, 5min) 0.83 Wh Baseline
Laptop Usage (5min) 2.5 Wh 2.5 min of 60W bulb
ChatGPT Query (Complex) 4.5 Wh 4.5 min of 60W bulb
Microwave (30sec) 25 Wh 25 min of 60W bulb
US Household (1min) 35 Wh 35 min of 60W bulb

Visual Comparison

Energy Efficiency Tips

  • Batch multiple AI queries together
  • Use simpler prompts when possible
  • Choose energy-efficient AI providers
  • Consider local processing for sensitive data

CO₂ Emissions: AI Models vs Real Life (2024)

Training vs Usage Emissions

AI Model Training

One-time massive emissions

25-700+ tonnes CO₂

Daily Usage

Small per-query emissions

0.001-0.01g CO₂/query

GPT-3 Training

502 tonnes CO₂

=8 years of car driving

BLOOM Training

25 tonnes CO₂

=25 round-trip flights NYC-SF

Daily ChatGPT Use

~0.1 kg CO₂/year

=2 hours of home heating

Carbon Footprint Context

While AI training has high upfront emissions, daily usage is relatively low. A year of heavy ChatGPT use equals about one car trip to the grocery store.

Own Your Data — Or Lose Your Privacy

81% of people feel little control over their data

Source: Cisco Privacy Report, 2023

Example: Zoom Data Usage

Zoom used calls to train AI (2023) - highlighting the importance of checking data usage policies.

DO:

  • Turn OFF "AI training" in apps
  • Keep mic/camera/location OFF when not needed
  • Review privacy settings monthly
  • Use privacy-focused alternatives

DON'T:

  • Keep mic/camera/location always ON
  • Accept default data sharing settings
  • Ignore privacy policy updates
  • Auto-sync sensitive data

Privacy Protection Tools

ProtonMail

Privacy-first email with end-to-end encryption

Visit Site

Signal

Secure messaging with disappearing messages

Visit Site

ProtonVPN

Encrypts internet traffic, hides location

Visit Site

DuckDuckGo

Search engine that doesn't store history

Visit Site

Fight Bias - Don't Let AI's Bias Decide Your Fate

AI misidentifies Black faces 100x more often than white faces

Source: NIST, 2019

Error rate disparity across demographic groups

Real Example

Apple Watch failed heart readings on darker skin (2021) → fixed in 2022

DO:

  • Request human review for rejections
  • Use IBM AI Fairness 360 (free toolkit)
  • Ask for reasoning behind AI decisions
  • Test AI with diverse datasets

DON'T:

  • Blindly accept AI decisions
  • Skip asking for explanations
  • Ignore demographic disparities
  • Use AI for high-stakes decisions without oversight

Combat Misinformation - Fact-Check AI Before You Share

AI fake news spreads 6× faster than human-written news

Source: MIT, 2023

Legal Consequence

Lawyer fined $5K for citing fake ChatGPT case. In 2023, a U.S. lawyer was fined $5K after citing fake legal cases generated by ChatGPT in a court filing.

BBC Verify

News verification team

Visit Site

Snopes

Fact-checking website

Visit Site

GPTZero

Detects AI-generated text

Visit Site

Weekly Ethical AI Checklist

Review app permissions and turn off unnecessary data sharing
Test one AI tool for bias using available detection tools
Verify AI-generated content before sharing on social media
Label any AI-generated work with proper attribution
Support privacy-first tools and ethical AI companies
Update passwords and security settings on AI platforms
Check energy consumption of your AI tool usage

Completion Progress

0 of 7 tasks completed this week

Your AI Ethics Action Toolkit

InVID

Verify videos/images for authenticity

Visit Tool

IBM AI Fairness 360

Check & reduce bias in AI

Visit Tool

Reality Defender

AI tool for detecting deepfakes

Visit Tool

FTC Complaint

Report harmful AI practices (U.S.)

Report Here

Sources & References

Research & Statistics

Pew Research Center (2024)

AI Adoption and Public Opinion Survey

View Report

PwC Global AI Study (2023)

Economic Impact Analysis - $15.7 trillion projection

View Study

Cisco Privacy Report (2023)

Consumer Data Control Perception Study

View Report

NIST Facial Recognition Study (2019)

Demographic Effects in Facial Recognition Technology

View Study

Environmental Impact

MIT Technology Review (2023)

AI Misinformation Spread Rate Analysis

View Article

Strubell et al. (2019)

Energy and Policy Considerations for Deep Learning in NLP

View Paper

University of Massachusetts Amherst (2019)

Carbon Footprint of Training Large Language Models

View Paper

Tools & Resources

IBM AI Fairness 360

Access Tool

Reality Defender

Access Tool

InVID Project

Access Tool

ProtonVPN

Access Tool

Signal Messenger

Access Tool

Data Accuracy Note

All statistics and figures cited in this guide are sourced from peer-reviewed research and reputable organizations. Energy consumption estimates are based on available public data and may vary depending on model architecture, hardware, and usage patterns. For the most current information, please refer to the original sources linked above.

Take Action Today!

Privacy First

Turn OFF data sharing in one app today

Be Transparent

Label one AI-generated post #MadeWithAI

Fight Bias

Test an AI tool with AI Fairness 360