A practical description of data and AI ‘ethics’ usually starts with a set of foundational moral principles, from which an organisation’s actions are derived. The first question then for a council should be, what are these moral principles? The second question, explored in the next capability, is how to adopt these principles. This should be the initial starting point that all the other capabilities serve.
How did we decide what principles to include?
In our initial framework, LOTI suggested that councils should the following common principles: Creates Public Value, Transparency, Accountability, Privacy, Fairness and Safety. These principles were chosen because LOTI research found them to be both relevant for local authorities and almost universally shared in other principles documents from around the world (such as from Camden, the Government Digital Service and Central Digital and Data oFfice, or the OECD here or here). You can read the original research here.
We no longer recommend these exact principles. To avoid the proliferation of different sets of principles, and to promote harmony between different regulations, LOTI now suggests that councils adopt the five regulatory principles set out in the UK Government’s “Pro-innovation approach to AI regulation”, and continue the principle of Privacy. These principles are extremely similar to the original set of principles. We suggest using the new ones because they will be the most pervasive principles, and are these are the regulatory principles that UK government has suggested regulators implement regulation around in their guidance.
The two other sets of guidance that councils may consider from central government are the GDS/CDDO Data Ethics Framework principles, and or the principles from the ‘Model for Responsible Innovation‘ created by the Department for Science Technology and Innovation. These can both be useful, but because the AI Regulation principles will inform regulations themselves, LOTI suggests using these principles.
However, there are other principles that LOTI thinks councils may want to consider including. The first extra principle that councils may want to include is ‘Creates Positive Social Outcomes’. LOTI suggests this because sometimes ethical principles can read as negatives to avoid. However, sometimes councils may be compelled to act to deliver a positive social outcome, for example with an action to improve the health or wellbeing of a resident or residents. Sometimes, failing to act or use technology could be considered unethical. (This principle is accounted for when complying with UK GDPR or in your Equalities Duty because the organisation must understand and detail the positive reason for using data or AI.)
The other principles that are worth considering for councils are any principles that they have as part of existing corporate policies. For example, some councils have principles, priorities or ways of working that are often detailed in a Council Plan or Local Plan. Within London you can see these in every council, but just to highlight a few examples from inner- and outer- London, and under different political leaderships, you can see examples from Ealing, Hammersmith & Fulham, Harrow, or Wandsworth. Common principles and themes that councils may want to include from these plans are around climate and sustainability, fairness and equality, and collaboration and co-production with residents.
1. Creates Positive Social Value
The first and default and fundamental principle for government in a democracy, government should always be trying to creating public value, delivering positive outcomes for residents and the people they serve. Read more here about the LOTI Outcomes-Based Methodology which we suggest should guide data projects.
2. Privacy
Individuals (and occasionally groups of people or communities) have a right and expectation of privacy – that the government will do their utmost to protect and guarantee their privacy even when innovating in new ways with data.
3. Safety, security & robustness
In some cases, advanced data uses or other technologies may even cause harm or compromise resident safety. To ensure this is happening, authorities need to ensure that data or AI systems function in a robust, secure and safe way throughout their life-cycle, and risks should be continually identified, assessed and managed.
4. Appropriate transparency and explainability
To foster trust and enable accountability, local authorities need to be open about the technology they use, including why and how they use it, what decisions are made with it, and who is making those decisions. This will require being transparent wherever possible, and being able to explain to residents (with differing levels of data and AI literacy) how they are using this technology.
5. Fairness
New uses of data (and the decisions made with them) can very easily create new or amplify existing biases in how our society, organisations or data projects have functioned previously. To counter this, authorities must develop consistent standards and approaches to know how to eliminate this, to ensure that the benefits of data are felt by all, including (and especially) those historically marginalised by the benefits of innovation.
6. Accountability and Governance
As with all decisions and actions taken, residents will expect to hold a local authority accountable for any decisions made about them with data or by using AI. Local authorities have a responsibility to create governance mechanisms that ensure they are accountable to the public.
7. Contestability and Redress
Whenever a council makes a decision using data or AI, residents have rights (for example within UK GDPR) to challenge and understand how that decision was made. If the decision was incorrect, residents will also have a right to redress, and this should be considered when designing services that use the technology. This might be particularly challenging with AI systems, which may be pre-developed in an opaque way, preventing a council from understanding exactly how the output was arrived at.
Camden created a Data Charter with its residents, which identified the following principles (which map onto those suggested by LOTI):
1. Build trust through transparency
2. Provide accountability and oversight
3. Make sure data is secure, safe and ethical
4. Make sure data is used for public good and be mindful of residents’ data
5. Be beneficial for all by using an outcomes-based approach
6. Be clear about how we use residents’ data
7. Protect individuals’ rights and privacy
