Responsible AI

Introduction to GPTs

Jasmin Ludolf

Data Science and AI Content Developer, DataCamp

Responsible AI

 

The practice of developing and using AI systems in a way that benefits society while minimizing the risk of negative consequences

1 https://www.iso.org/artificial-intelligence/responsible-ai-ethics
Introduction to GPTs

Responsible AI

 

The practice of developing and using AI systems in a way that benefits society while minimizing the risk of negative consequences

 

  • Everyone plays a part in safe AI use
  • We can't blame the tool!

Illustration of a wooden legal hammer

1 https://www.bbc.co.uk/news/world-us-canada-65735769
Introduction to GPTs

Considerations

Lawful

Introduction to GPTs

Considerations

Fairness

Introduction to GPTs

Considerations

Transparency

Introduction to GPTs

Considerations

Diversity

Introduction to GPTs

Considerations

Inclusivity

Introduction to GPTs

Considerations

Accountability

Introduction to GPTs

Considerations

Privacy

Introduction to GPTs

Considerations

Security

Introduction to GPTs

User considerations

  • Behavior complies with laws and regulations
    • Different depending on location and industry
  • Treat everyone equally without discrimination
  • Understand how the tool works
  • Respect individual rights
  • Protect business data
  • Applies to prompts and outputs

Person sitting at a desk with a laptop reading documents

Introduction to GPTs

Check the policy

  • Code of conduct or policy
  • How to use it
  • Data expectations

  • 🚨 Follow these policies

Guidelines on:

  • Financial data
  • Personal data
  • Proprietary data

Policies may differ!

A person holding a long scroll document with policy written at the top

Introduction to GPTs

General input considerations

  • Personally identifiable information:

    • ID numbers
    • Addresses
    • Medical
    • Banking
  • Proprietary information:

    • Confidential strategies
    • Legal
    • Financial

GPT robot holding a stop sign

Introduction to GPTs

General input considerations

A person sitting with a laptop and a thought bubble containing the GPT robot holding the stop sign.

                                              - Provider using the data

                                              - Vulnerable to attacks

Introduction to GPTs

General output considerations

  • Bias: unfairly favors one outcome
  • Context tracking: losing track of the conversation
  • Hallucination: confident incorrect answer

 

No 100% guarantee

 

Our responsibility

GPT bot showing an incorrect answer saying the sky is green

Introduction to GPTs

Let's practice!

Introduction to GPTs

Preparing Video For Download...