March 23, 2022
Ethical AI Part 10: Trust
We do what we promise. And we take responsibility for our algorithms.
All the ethical intentions in the world are worthless without earned trust.
Despite the growing importance and reliance on AI technology, earning consumers’ trust is one of the biggest challenges facing the broader adoption of AI. Factors contributing to the lack of trust in AI systems include concerns surrounding privacy, bias and security to a general misunderstanding of how AI systems function. Levels of trust in AI also vary significantly around the world, suggesting that socio-economic and political factors also need to be considered when asking consumers to invest their trust in something they can’t touch or see.
Trust is an essential part of AI. Without trust, the decision-making tools powering AI systems cannot deliver the intended results. Honesty and transparency are critical contributors to trust in AI, and at Utopia Analytics, not only do we consider trust, honesty and reliability as fundamental values of our business, but as cornerstones of our ethical AI services, too. Trust is imbued into every part of our business and how we work. After all, if our customers and partners can’t trust us to act ethically, how can we ask the same of them?
The European Commission’s Ethics Guidelines for Trustworthy Artificial Intelligence lists seven essential requirements that AI systems need to be trustworthy, all of which must be continually evaluated and assessed through the AI’s lifecycle. These are:
- Human agency and oversight
- Technical robustness and safety
- Privacy and data governance
- Diversity, non-discrimination and fairness
- Societal and environmental wellbeing
We’ve discussed many of these points, such as accountability, transparency and human agency, in other blogs. AI models are only as powerful and trustworthy as the training data they are fed, which is why we place such an importance on trust and transparency at an operational level at Utopia. AI must be built and deployed responsibly so we can maximise its benefits to humanity, and this is also why we believe the UDHR is the best foundation to safeguard the ethical use of our AI.
Whether it’s AI moderation or AI-powered claims processing, all of Utopia’s AI products are engineered with the highest standards in mind. Humans always define the task that AI is applied to, ensuring we make the proper ethical considerations before the technology is deployed. To ensure neutrality in decision-making, Utopia also never defines a platform’s moderation policy; we leave it up to our customers to decide.
As AI systems’ role in our everyday lives continues to grow, providers and users of these technologies must embed trust at the heart of their operations. Otherwise, issues around consumer attitudes to AI stemming from lack of trust can be the most significant barrier to consumers experiencing the benefits that AI systems were created to deliver.
Building trust into our work begins with the careful screening of prospective customers as part of our sales process, as we don’t allow our AI products to be used in situations that may compromise human rights or impede freedom of expression in any way.
More generally, our staff stick to the facts, stick to the timelines, and only make promises that we can keep. This ensures our customers (and their end-users) can always trust us to stick to the same ethical guidelines that we expect them to follow.
We aim to continue this mantra once our work is delivered and the system is up and running. Our products and AI models are built to last and scale as your business grows. But staff are always on hand for support and maintenance issues with no additional fees.
That’s AIaaS (AI as a service). That’s responsibility in action. That’s the formula of trust, and a key cornerstone of our Ethical AI.
Ethical AI part 1: Time to talk about responsibility
Ethical AI part 2: Powerful tools need a solid base
Ethical AI part 3: Human rights breach as grounds for termination
Ethical AI part 4: AI moderation and freedom of expression
Ethical AI part 5: Values in order
Ethical AI part 6: No prejudice
Ethical AI part 7: Honesty
Ethical AI part 8: Equality
Ethical AI part 9: Safety