AI Usage and Ethical Considerations Policy
Effective Date: 02/12/2024
Last Updated: 02/12/2024 Rev: 1.1
1. Introduction
At Riskoa, we are committed to leveraging artificial intelligence (AI) responsibly and transparently to enhance the quality, efficiency, and accessibility of our lifecycle assessment services. This policy details the role of AI within our offerings, the shared responsibilities of Riskoa and its customers, and the mechanisms in place to address limitations and improve outcomes.
The Emvide platform, powered by advanced AI, transforms lifecycle assessments by delivering rapid, data-driven insights. However, AI is a computational tool that can make mistakes, and its outputs are influenced by the quality and completeness of data inputs. This policy acknowledges these limitations while outlining our continuous commitment to improvement.
This policy applies to all AI-powered services provided by Riskoa, including the Emvide platform, the LCA Factory service, and any future proprietary AI systems.
2. AI Capabilities and Role in Emvide
The AI-driven functionality of Emvide underpins its ability to support and accelerate lifecycle assessment processes. Specific capabilities include:
Semantic Search: The AI identifies emissions factors and lifecycle inventory data from global datasets, reducing the complexity of manual searches.
Data Decomposition: Complex product information is segmented into analysable components, facilitating precise assessment.
Scenario Modelling: The platform generates sensitivity analyses and explores alternative pathways for reducing environmental impacts.
Report Automation: Product Carbon Footprints (PCFs), Lifecycle Assessments (LCAs), and Environmental Product Declarations (EPDs) are generated with consistency, speed, and reliability.
2.1 Use of Third-Party AI Services
To deliver advanced functionalities, Emvide integrates with third-party AI services. These services provide specialised capabilities, such as enhanced semantic searches and data enrichment, essential to the performance of the platform.
Data Sharing with Third-Party AI Services:
Portions of customer data may be shared with these services under the following conditions:
Only the minimum data required to execute specific tasks is shared.
Data is anonymised wherever feasible to protect customer confidentiality.
Third-party providers adhere to stringent security standards in compliance with applicable regulations.
Riskoa actively monitors the performance of third-party services and ensures contractual safeguards are in place to mitigate risks.
2.2 Future Development: Proprietary AI Systems
Riskoa is committed to advancing its AI capabilities through the development of a proprietary Large Language Model (LLM). This system will be tailored specifically for lifecycle assessments and designed to:
Enhance functionality and accuracy.
Improve data security and customer privacy.
Offer customisation options for clients requiring bespoke AI solutions.
Clients interested in dedicated AI deployments are encouraged to contact Riskoa to discuss customisation opportunities.
3. Acknowledging Limitations of AI
AI is a powerful tool, but it can make mistakes due to the following factors:
Data Dependency: Outputs are only as accurate as the data provided. Missing, incomplete, or inaccurate inputs can affect results.
Assumptions: In cases where data is ambiguous, the AI may make reasonable assumptions that might not align with customer expectations.
External Dependencies: Performance may be impacted by interruptions or limitations in third-party AI services.
For the Emvide platform, users are responsible for reviewing outputs and verifying their applicability. In contrast, for LCA Factory, Riskoa assumes responsibility for delivering validated and compliant outputs as part of the service.
4. Shared Responsibility
The delivery of accurate and reliable outputs is a collaborative effort between Riskoa and its customers.
4.1 Customer Responsibilities
Customers are expected to:
Provide accurate, complete, and well-structured input data.
Review AI-generated outputs and promptly report discrepancies.
Use the platform in accordance with guidelines provided by Riskoa.
4.2 Riskoa’s Responsibilities
Riskoa is committed to:
Conducting thorough quality assurance for deliverables, particularly in LCA Factory services.
Regularly auditing and refining AI systems to improve accuracy and reduce errors.
Responding promptly to customer feedback and taking corrective actions as necessary.
5. Feedback, Resolution, and Remuneration Policy
Riskoa is committed to customer satisfaction and continuous improvement. To address perceived errors in the Emvide platform or related services, we have implemented a structured Report, Resolve, Improve, and Remunerate framework.
5.1 Reporting Errors
To enable timely investigation and resolution, customers must report perceived errors within 24 hours of occurrence. Reports submitted outside this timeframe may not be addressed, and Riskoa reserves the right to refuse resolution or remediation in such cases.
How to Report Errors:
Customers must submit the following information through Riskoa’s designated support channel:
Detailed Description: A clear explanation of the issue, including the steps taken that led to the error.
Supporting Evidence: Relevant data, screenshots, or any other documentation to substantiate the claim.
Reports that lack sufficient detail may result in delays or an inability to process the issue effectively.
5.2 Resolution Process
Once an error report is submitted, Riskoa will follow a structured process:
Acknowledgment:
Reports will be acknowledged within one (1) business day of receipt.
Investigation:
Riskoa’s product development and support teams will thoroughly review the report to determine the root cause of the issue.
This may involve re-evaluating system outputs, customer data inputs, and any applicable processes.
Resolution:
Customers will receive a formal response within five (5) business days of the report, which may include:
A summary of findings.
Corrective actions or an explanation if no error was found.
Recommendations for future data submissions (if applicable).
5.3 Remuneration Mechanism
In cases where Riskoa determines that the error was caused solely by the Emvide AI system, and the error significantly impacts the deliverable’s usability or compliance, promotional credits may be offered as a goodwill gesture.
Strict Conditions for Remuneration:
Remuneration is granted at Riskoa’s sole discretion on a case-by-case basis.
Legal Obligation Disclaimer: Riskoa is not legally obligated to provide remuneration. It is offered as part of our commitment to fostering a learning environment and enhancing customer satisfaction.
Eligibility Criteria for Credits:
The error must be directly attributable to the Emvide AI system.
The issue must materially impact the deliverable’s accuracy, compliance, or usability.
Remuneration will not be issued if the issue is found to fall under excluded circumstances (see below).
Exclusions from Remuneration:
Riskoa will not issue credits for errors resulting from:
Customer Input Errors:
Incomplete, inaccurate, or poorly formatted data provided by the customer.
Improper Use of the Platform:
Deviation from guidelines or misuse of the Emvide system.
Exceeding Submission Deadlines:
Failure to report errors within the required 24-hour window.
External Dependencies:
Issues caused by third-party AI services, infrastructure outages, or network failures beyond Riskoa’s control.
Scope Changes:
Errors or discrepancies arising from customer-initiated changes to the scope of the deliverable after project commencement.
Assumptions in Data Interpretation:
Instances where assumptions were made due to incomplete or ambiguous data provided by the customer.
Customer Preferences:
Situations where outputs do not align with customer preferences or expectations unrelated to compliance or usability.
5.4 Remuneration Credits
When issued, remuneration will be provided as promotional credits applicable to future use of the Emvide platform or related services. The value and quantity of credits are determined by the following factors:
Severity of the Error: Extent of impact on the deliverable.
Scope of the Project: Size and complexity of the affected deliverable.
Effort to Resolve: Resources and time required to address the issue.
Promotional credits have no cash value, are non-transferable, and must be used within the validity period specified by Riskoa.
Riskoa is committed to addressing errors in good faith and learning from reported issues to improve the Emvide platform. However, we emphasise that remuneration is offered voluntarily as part of our customer-centric philosophy and not as an admission of liability or legal obligation.
6. Continuous Learning and Improvement
Riskoa’s approach to AI is guided by continuous learning and customer feedback. Key initiatives include:
Proactive Audits: Regular reviews of AI models to improve performance.
Feedback Integration: Incorporating customer insights into system updates.
Future Innovations: Expanding AI capabilities, including proprietary solutions like the LLM.
7. Ethical AI Principles
Riskoa adheres to ethical AI practices to promote fairness, accountability, and transparency. These include:
Ensuring that AI outputs are free from bias and discrimination.
Providing clear explanations of AI capabilities and limitations.
Taking accountability for AI-related errors and their resolution.
8. Data Privacy and Security
The security of customer data is central to our AI operations. Riskoa ensures that:
Shared data is minimised and anonymised whenever possible.
Encryption safeguards data during storage and transmission.
Data shared with third-party providers complies with stringent security standards.
For further details, refer to our Privacy and Data Protection Policy.
9. Contact Information
10. Updates to the Policy
This policy will be updated as necessary to reflect advancements in AI technology, changes in regulatory requirements, or customer feedback. Customers will be notified of significant updates through the Trust Centre or direct communication.
Last updated