> For the complete documentation index, see [llms.txt](https://ersilia.gitbook.io/ersilia-book/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ersilia.gitbook.io/ersilia-book/about-us/ai-policy.md).
# AI Policy
### Purpose of the Policy
An AI policy is essential to ensure that artificial intelligence tools are developed and deployed responsibly, ethically, and in alignment with organizational values. This policy establishes clear guidelines for the use of AI to protect stakeholders, maintain trust, and promote equitable outcomes.
### AI Use
AI tools are used by the Ersilia Open Source Initiative (Ersilia) to support research and healthcare decisions, with direct interaction between beneficiaries—including researchers, healthcare workers, and communities in the Global South—and the AI systems. Scientific and research data, such as molecular structures, disease data, and experimental results, as well as personal data including names and contact information of users and partners, are collected and processed through these tools.
The organization's AI applications are designed to advance scientific discovery and improve health outcomes in resource-limited settings.
### Interactions with Beneficiaries
AI tools are deployed directly to beneficiaries who use them for research and healthcare decision-making. These interactions require careful attention to usability, accessibility, and the provision of clear guidance on how to interpret AI-generated outputs. Users receive training to understand the capabilities and limitations of the AI tools to ensure informed and appropriate use in their work.
### Data Collection Practices
Explicit consent is obtained from all users before collecting or processing their data, and clear information is provided about what data is collected and how it is used (see our [Privacy Notice](/ersilia-book/about-us/ersilia-privacy-notice.md)). Data privacy and security are prioritized through transparent communication with users regarding data handling practices. All data collection activities are conducted in accordance with ethical standards to protect the privacy and rights of researchers, healthcare workers, and partner organizations.
### Ethical Risks and Concerns
Privacy and security of research data and user information are paramount, requiring robust safeguards to prevent unauthorized access or misuse. Transparency in how AI models make predictions or recommendations is essential to build trust and enable users to critically evaluate AI outputs. Equitable access to AI tools across different regions and resource levels must be ensured to prevent the exclusion of underserved communities and to promote global health equity. These ethical considerations guide the development, deployment, and ongoing evaluation of all AI systems.
### Accountability for AI Decisions
Shared accountability is maintained between Ersilia staff who develop and maintain the AI models and the researchers or healthcare workers who use the AI outputs in their decision-making processes. Ersilia staff are responsible for ensuring that AI models are accurate, reliable, and appropriately documented, while end users are responsible for applying AI-generated insights appropriately within their professional contexts. This collaborative accountability framework recognizes that both developers and users play critical roles in ensuring ethical and effective AI deployment.
### Bias Detection and Prevention
AI models are tested across diverse datasets to ensure broad applicability. While third-party tools are used with the assumption that they are unbiased, ongoing vigilance is required to monitor for unexpected biases that may emerge in practice.
### Transparency in AI Decisions
Clear documentation is provided explaining how AI models work, their limitations, and the appropriate contexts for their use. AI models are made open source whenever possible, allowing users to inspect, understand, and validate the underlying algorithms and methodologies. Users are supported in understanding the basis for AI-generated predictions through accessible explanations and educational resources. This commitment to transparency enables informed decision-making and fosters trust in the AI tools.
### Community Feedback and Reporting
Formal partnerships with organizations in the Global South provide ongoing feedback on the effectiveness and appropriateness of AI tools in diverse contexts. Open communication channels are maintained to facilitate user input and enable rapid identification of issues or concerns. Feedback from the research community and beneficiaries is collected at the end of collaborative projects and used to improve AI tools and ensure they meet the evolving needs of users.
### Third-Party AI Tools and Accountability
Third-party AI tools are used in accordance with their existing terms of service, and reliance is placed on vendors' stated commitments to ethical AI practices. While this approach provides operational efficiency, it requires ongoing monitoring to ensure that third-party tools continue to align with Ersilia's ethical standards and mission. Any concerns regarding third-party tools will be promptly investigated and addressed through communication with vendors or by seeking alternative solutions.
### Staff Training and Awareness
Staff members rely on their existing knowledge and expertise regarding AI ethics and best practices, with the expectation that they maintain awareness of evolving standards in the field. While formal training programs are not currently in place, staff are encouraged to engage in professional development opportunities and to share knowledge within the organization. As AI technologies and ethical considerations evolve, the organization may develop more structured training initiatives to ensure consistent understanding of responsible AI use across all team members.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://ersilia.gitbook.io/ersilia-book/about-us/ai-policy.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.