Ethical and responsible use of GenAI

The use of Generative AI (GenAI) comes with risks and ethical considerations. As you explore GenAI’s potential, it is important to understand some of the critical aspects surrounding its use. For a deeper look at any of these topics, see the relevant section of the GSLS GenAI Tutorials! 

1. Data ownership
2. Data security
3. Diversity and biases
4. Guarding against data corruption and LLM Grooming
5. Consent, Transparency and Accountability 
6. Equity
7. Environmental impact
    

1. Data Ownership:

Ownership of the data used to train GenAI tools is legally contested and still being tested in court. Copyright generally stays with the original creators or publishers, not the AI providers. Whether training on that material is legal can vary by jurisdiction and remains disputed. See Module 1 (1.9) for more. 

2. Data Security:

GenAI tools do not manage data storage or control access to specific datasets. They operate solely on pre-trained models and lack the capability to implement security measures for storage or restrict data access. To enhance security, do not place personal data or unpublished data into unsecure platforms, and consider disabling data training in the settings. See Module 2 (2.4) for full guidance. 

3. Diversity and Biases:

GenAI tools generate content based on the patterns in the data they were trained on. This data tends to exclude diverse perspectives, which can produce biased responses. It is crucial for users to critically assess and address these factors when using these tools. See Module 1 (1.9-1.10) for more on training data and embedded bias. 

4. Guarding Against Data Corruption and LLM Grooming:

GenAI tools can sometimes detect inconsistencies in input data, which may suggest potential tampering, including training-date manipulation such as LLM grooming. However, they do not have the capability to actively prevent or respond to tampering. Always check for accuracy. See Module 1 (1.10) for more about this. 

5. Consent, Transparency, and Accountability:

GenAI tools do not manage informed consent, verify sources, or take accountability for their outputs, these all remain with the user. When using GenAI you must ensure any required consent was properly obtained; when using it in your work, you must be transparent; and you must always verify accuracy. These responsibilities are also part of your obligations under Scientific Integrity.

6. Equity:

Many of the GenAI tools require paid subscriptions. This means that not all students have equal access, putting them at an unfair disadvantage. Therefore, students will not be asked to use paid versions of these tools by teachers or supervisors. See Module 2 (2.2) for more. 

7. Environmental Impact:

GenAI tools do not enforce accountability or integrity. This is a commitment users must make when using these tools. Always check for accuracy.GenAI tools, along with everyday online activities such as sending emails and conducting internet searches, depend on data centres that consume large amounts of energy and water, particularly for cooling servers. Therefore, it is important to adopt a conscious and responsible approach to digital engagements, ensuring our contribution towards the sustainability of our digital environment. See Module 2 (2.5-2.7) for more.