Public sector organisations looking to use AI need to pay careful attention to data quality, fairness, accountability and privacy, according to new guidance.
The Government Digital Service and Office of Artificial Intelligence have published the guide to help organisations understand the technology and make the most of the opportunities.
It comes alongside the publication of the Technology Innovation Strategy, which includes an emphasis on the importance of new technologies, and the unveiling of the new Spark procurement framework for the sector.
The guidance includes a section on considerations for using AI to meet user needs that includes the warning that the success of a project depends heavily on data quality, and that an organisation should assess the state of its data for factors such as accuracy, completeness, timeliness, sufficiency and consistency.
It also places a strong emphasis on ethics and safety, saying that an AI system should be implemented responsibly and without bias, that there should be mechanisms for the accountability of designers and implementers, and full compliance with the General Data Protection Regulation.
Transparency and costs
In addition, its operation should be transparent and explainable, and there should be a proper consideration of the costs of building, running and maintaining an AI infrastructure and training staff to use it.
The document also includes a warning that AI is not a general purpose solution for every problem, and that current applications have strong limitations. They cannot be imaginative, perform well without a large quantity of relevant, high quality data, or infer additional context if the information is not present in the data.
It says that simpler solutions can often be more appropriate, citing the example of a digital form working more effectively for some purposes than optical character recognition technology to extract information from documents scans.
“You’ll need to investigate alternative mature technology solutions thoroughly to check if this is the case,” the document says.
It comes with a number of related guides, including how to assess if AI is the right solution, planning and preparing for an implementation, managing an AI project and the National Cyber Security Centre’s guidance for assessing intelligent tools for cyber security.
Digital Secretary Jeremy Wright said: “Artificial intelligence is already having a positive impact across society, from improving fraud detection to better and quicker diagnoses of medical conditions.
“The UK Government has already been recognised as world leading in its readiness for AI and we continue to push leaders across the public sector to recognise its impact in delivering more personalised and efficient experiences.
“Our newly appointed AI Council of industry experts will boost the growth and use of AI in the UK further by helping us to realise its full potential.”
Image from iStock