Introduction
AI and machine learning is becoming an increasingly important part of daily lives for many since the invention of ChatGPT and consumer AI products. The possibilities are great for improved productivity with information at our fingertips, but, as with any instrument or tool there is risk of bias, drift or error. Large Language Models (LLMs) work by predicting the next token (part of a word), but with a context potentially in the millions of tokens even very precise predictions with only small errors can accumulate into large drifts. Imagine travelling 1cm perfectly straight at a time with 99.99% accuracy in direction of travel. However the combined chances of error mean one significant error in the chain could send the entire journey in a different direction. Hence just as we take responsible measures to ensure our laboratory instruments are giving the right answers with proficiency testing, calibrations, staff training to establish uncertainty, the same measures should also be taken when using any machine learning algorithms too. Every test has a margin of error, as long as it’s understood and managed and customers are informed then AI can be a tool for good. This policy outlines how for when failure is not an option we at Oil Analysis Laboratories use AI and machine learning in all its forms in a responsible way in a rapidly changing industry landscape.
Our Commitment
We recognise that artificial intelligence brings both opportunity and responsibility — and we are committed to using it transparently, ethically, and with full accountability to our clients, employees, and the wider community.
Our goal is to ensure every machine learning or AI-driven system we use improves accuracy, safety, and efficiency without compromising fairness, data integrity, or human oversight.
Purpose
This policy explains how we govern, monitor, and continually improve all AI systems used within our laboratory and digital platforms.
Principles We Follow
Transparency
We clearly communicate when and how AI is used within our services — from automated data interpretation to predictive maintenance insights.
Where automated models assist with diagnostics and have done for decades in labs, final decisions always remain under qualified human review.
Accountability
We take full responsibility for every AI-supported outcome delivered to our customers.
All AI systems operate under controlled procedures, version control, and extensive testing before release with re-evaluation regularly.
Fairness & Non-Discrimination
Our models are trained and tested to avoid bias or unfair outcomes.
We regularly review datasets and algorithms to ensure results remain objective, representative, and technically sound.
Human Oversight
AI is a decision-support tool, never a substitute for expert judgement and we encourage our customers the same when making million pound decisions off the back of a quick chat with AI.
Data Protection & Privacy
All data used in AI training or analysis is handled securely in accordance with relevant data protection and cyber security regulations
Sensitive or client-identifiable information is both encrypted to ensure we are training on what we are doing and never on our customers.
We never have, never considered and never will consider selling, exploiting or giving away you the customers data. We are in the market of selling samples not data and all clients can rest assured we value their trust more than anything and will never endanger the relationship.
Safety & Reliability
Machine learning systems undergo rigorous testing to confirm accuracy, repeatability, and robustness.
We use continuous monitoring to detect model drift, anomalies, or unexpected results.
Continuous Improvement
We treat machine learning models as we do any other laboratory process which is reviewed, retrained, and refined as our laboratory and clients evolve.
Regular audits, impact assessments, and staff training ensure our approach remains responsible and compliant.
Governance and Oversight
Our AI Governance Committee including board and senior management oversees all major AI developments and applications.
Together, they approve new AI tools, review risks, and ensure controls align with the most stringent of AI safety standards.
How We Use AI
AI is not just Large Language Models, although we do have customer support chat bots too.
We use AI to augment what our lab staff do. For instance we use visual models to augment and support our staff. When a subjective interpretation such as is an oil dark or light is made, the AI assists in highlighting samples that may require a second opinion before release. When it comes to diagnostics AI supports ways of better phrasing a comment and highlights points the diagnostician may have missed just before releasing the report or even when the customer has given some sample data that can’t be correct for the sample.
Every deployment is risk-assessed, validated, and traceable. No AI system operates unsupervised in safety-critical or compliance-sensitive contexts and instead the software helps catch issues for secondary human evaluation before release to the customer.
Ethical Standards
We adhere to the following ethical commitments:
AI will never be used to deceive, mislead, or manipulate outcomes. Decisions that could impact people, safety, or business reputation will always include human validation. We maintain a responsible innovation approach — balancing scientific advancement with ethical foresight.
Reporting and Feedback
We welcome scrutiny and dialogue around our use of AI. If you have concerns, questions, or suggestions about how we use automation or artificial intelligence, please contact the lab by email using the contact buttons on our website.
All concerns are reviewed confidentially by our governance team.
Review & Continuous Improvement
This policy is reviewed annually or whenever there are significant technological or regulatory developments in AI governance and updated on our website.
Statement from our founder
“Artificial intelligence amplifies what we already do best — deliver precise, timely, and scientifically robust results.
But we believe responsibility comes first.
That’s why every AI-enabled decision at Oil Analysis Laboratories remains transparent, ethical, and under human control.”
— Adam Cutler, Managing Director
