Use AI Responsibly
Overview
- Responsible AI: The principle of developing and using AI ethically with the intent of benefiting people and society, while avoiding harm.
- AI user: Someone who leverages AI to complete a personal or professional task.
- AI operates as a complement to uniquely human skills (creativity, logic, compassion, critical reasoning, and contextual understanding).
- AI is not perfect and has limitations; it cannot make complex, higher-level decisions (e.g., personalized performance feedback, hiring judgments, therapy).
Note: To ensure people are treated fairly and respectfully, AI users must be aware of the limitations of AI tools and commit to using them ethically.
Recognizing Bias in AI
AI models are trained on data created by humans and reflect the values of their designers. Consequently, they are not intrinsically value-neutral and can produce inaccurate or biased results.
- Systemic bias: A tendency upheld by institutions that favors or disadvantages certain outcomes or groups. This exists within societal systems like healthcare, law, education, and politics.
- Data bias: A circumstance in which systemic errors or prejudices lead to unfair or inaccurate information, resulting in biased outputs. If training data lacks diversity, the AI model's output will reflect that exclusion.
Identifying AI Harms
If used without human intervention or critical thinking, AI can reinforce systemic bias, leading to an unfair distribution of resources, the perpetuation of dangerous stereotypes, or the reinforcement of ongoing power dynamics.
- Allocative harm: A wrongdoing that occurs when an AI system's use or behavior withholds opportunities or resources or information in domains that affect a person's wellbeing (e.g., an AI screening tool denying housing due to mistaken identity).
- Quality-of-service harm: A circumstance in which AI tools do not perform as well for certain groups of people based on their identity (e.g., speech recognition technology failing to understand speech patterns of people with disabilities).
- Representational harm: An AI tool's reinforcement of the subordination of social groups based on their identities (e.g., a translation app associating gender-specific translations to certain professions based on assumptions).
- Social system harm: Macro-level societal effects that amplify existing class, power, or privilege disparities or cause physical harm as a result of the development or use of AI tools.
- Deepfakes: AI-generated fake photos or videos of real people saying or doing things that they did not do. The spread of deepfakes causes large-scale disinformation.
- Interpersonal harm: The use of technology to create a disadvantage to certain people that negatively affects their relationships with others or causes a loss of one's sense of self and agency (e.g., misusing private information to lock someone out of an online account).
The Path to Working in Responsible AI
Mitigating bias and harms in AI requires active human involvement and critical thinking:
- Check the outputs of AI tools for accuracy.
- When ingesting new data, ensure it is inclusive and representative of different communities.
- Continually retrain and fine-tune models using diverse datasets.
- Provide feedback continuously to help developers improve AI systems.
Security and Privacy Risks of AI
As generative AI introduces new capabilities, it also presents new security risks for organizations and individuals.
- Privacy: The right for a user to have control over how their personal information and data are collected, stored, and used.
- Security: The act of safeguarding personal information and private data, and ensuring that the system is secure by preventing unauthorized access.
Best Practices for Privacy and Security
- Read the terms of use, privacy policy, and any associated risks before using an AI tool. Understand how the tool collects and uses data.
- Never input personal details (e.g., identity, budget details, email addresses) or confidential information into an AI prompt. This prevents sensitive data from being exposed during a security breach or data leak.
- Keep up to date on new advancements in AI to understand emerging risks. Consult trusted news sources, scholarly publications, and subject matter experts.