Peppermint Technology and their responsible approach to AI
With rapid advancements and improvements in AI over the last few years, textual and image generation have leapt forward at an unprecedented pace. “As a veteran technologist, I’ve never seen such swift progress”, says Mike Walker, Peppermint’s CITO. This evolution brings both incredible opportunities and significant responsibilities. As technology leaders and consumers of AI technologies, it is crucial for us to consider how these advancements can be used both positively and negatively.
Our primary focus at Peppermint Technology is to leverage AI and advanced technology in a way that is responsible and ethical. We have developed internal policies to guide responsible AI usage and are committed to enforcing these within our designs. We actively evangelize the importance of considering potential risks and impacts on people. The potential misuse of AI is a significant concern, and we are dedicated to addressing these challenges.
In this document, Mike Walker will outline Peppermint’s approach from a pillared perspective, highlighting the key principles guiding our responsible AI initiatives.
We also hope that others in the industry will reflect on these considerations and join us in fostering responsible AI development and deployment.
1. Fairness
We are committed to fairness throughout our AI product lifecycle. Our goal is to eliminate bias and prevent discrimination in our AI systems. By leveraging our strong partnership with Microsoft and other technical leaders, we enforce policies and activate features that enhance fairness. Our objective is to ensure our AI technologies operate justly and equitably.
2. Reliability and Safety
Reliability and safety are paramount. We utilise Azure OpenAI services features to integrate robust content safety filters and rigorously test our products. This helps us understand and mitigate potential impacts and unintended consequences. We continuously fine-tune these features to suit our industry needs, ensuring AI responses are well-informed and safe. User awareness regarding the use of AI in our applications is a top priority.
3. Privacy Protection and Compliance
Protecting customer privacy is fundamental, especially in the legal sector. We adhere to strict regulatory requirements, ensuring data is encrypted and never shared without consent. Our use of Microsoft’s OpenAI services involves de-registered content to prevent human interaction with sensitive data. We are proactive in safeguarding our customers’ and their clients’ data.
4. Security Measures
Security is a top priority in our AI initiatives. We leverage Microsoft’s industry-leading services to ensure encrypted data transmission and robust end to end controls. Server-side grounding minimises the risk of data being exposed over the public internet. Our commitment to security ensures that AI-generated information is accessible only to internal systems, enhancing data protection. No information is ever learned or added to a model.
5. Responsible AI Maturity
We are committed to advancing AI responsibly. By working with Microsoft and other industry leaders, we focus on maturing our AI processes both for internal use as well as what we deliver as product features. We continually assess and refine our AI policies, ensuring transparency and ethical governance. Our goal is to deliver secure and robust AI solutions that inspire confidence among our customers.
6. Inclusive Design
Inclusivity is a core principle in our AI product development. We design AI-driven solutions to be accessible and beneficial to all users, regardless of background or abilities. Our commitment to inclusiveness extends to our licensing policies, ensuring that our features are available without discrimination. Microsoft’s approach aligns with this by ensuring diverse data representation, mitigating biases through tools like Fairlearn, and emphasizing human-centric design. They prioritise transparency in AI decision-making and hire a diverse talent pool to minimize biases and create equitable AI technologies.
7. Transparency
We uphold transparency in all AI-related activities. AI-generated content within our products and documentation is clearly indicated. Users are provided with clear explanations and options to disable AI features if desired. Our transparent approach ensures users are fully informed about AI usage in our applications.
8. Accountability
We hold ourselves accountable for the ethical use of AI technologies. Clear lines of accountability are established within our development and deployment processes. We ensure thorough testing and ethical considerations in all AI initiatives, proactively addressing potential concerns.
Conclusion
At Peppermint Technology, we are dedicated to leading the way in responsible AI internally in our business systems as well as in the solutions we develop for our clients use. Our eight core principles reflect our commitment to fairness, safety, privacy, security, maturity, inclusivity, transparency, and accountability. By adhering to these principles, we aim to foster trust and confidence in our AI solutions, ensuring they benefit all stakeholders responsibly and ethically.
It is our responsibility as technology leaders to ensure that AI is developed and used ethically. This involves implementing robust safeguards, continuously monitoring and evaluating AI systems, and fostering a culture of transparency and accountability. By doing so, we can harness the power of AI to drive positive change while minimizing the risks associated with its misuse.
We must also engage in open and ongoing dialogue with stakeholders, including regulators, industry peers, and the public, to address the ethical challenges posed by AI. Collaboration and knowledge-sharing are essential to developing best practices and standards that promote the responsible use of AI.
Ultimately, our goal is to build AI systems that are not only powerful and efficient but also ethical and inclusive. By prioritising fairness, safety, and transparency, we can ensure that AI technologies enhance our lives and society in a positive and meaningful way.
