Public Sector
Automate and improve the efficiency of public services using unbiased and transparent machine learning models.
Public Sector in the age of AI
In the public sector, accountability and transparency in decision making are vital. Areas of government such as healthcare and policing are actively taking advantage of recent advances in machine learning that enable typical black-box approaches to become transparent and explainable. At ADSP, we use computer vision algorithms and AI responsibly and ethically to discover patterns and predict events that improve the delivery of public services.
Executive opinion
How can public services use AI to responsibly inform decision-making?
When deploying machine learning algorithms that impact public services, it is important to assess the potential for unintended bias in the data and model design. Embedding fairness metrics into the training process of the model, isn't optional - it's an essential part of assessing the validity and applicability of the algorithm, just as much as any accuracy metric. By fully understanding and auditing the data pipeline, public services can harness the power of AI in a way that is ethical and represents all areas of society equally.
David Foster
Founding Partner, ADSP
The Office for AI estimates that AI could add £232 billion to the UK’s economy by 2030, boosting productivity in some industries by 30%.
Artificial Intelligence and Public Standards Report, Feb 2020
High value AI projects for Public Sector
Healthcare
The impact of AI on healthcare is already well documented. At ADSP, we analyse large repositories of unlabelled and unstructured data to build deep learning models for computer vision tasks for AI-assisted diagnosis and mathematical optimisation models that improve logistical operations.
Polling Analysis (MRP)
Extrapolation of survey and polls is now more accurate than ever thanks to a technique known as MRP (Multilevel regression and post-stratification). We use this technique to deliver highly precise estimations by combining your survey results with census data, for an accurate reflection of results nationwide.
Predictive Policing
Existing policing measures can be enhanced through real-time monitoring of crimes and the use of graph networks to detect webs of criminal activity hidden in data. Predictive policing is already being used successfully to discover distinct patterns of criminal behaviour which are used to optimally deploy teams to where they are needed most.
