Spoiler, it may be a Common Data Model
The benefits of AI are being heralded in every quarter, with organisations rushing to embrace this next generation technology. But for organisations to see real benefit, they need to be discerning in the problems they are trying to solve. And to do that, they must understand how they harness the data on which AI relies.
Security and policing data challenges
All Government departments are different and will have unique problems, including legacy systems, demands on their services and regulatory pressures. All of which add an additional layer of complexity which can obscure some common data challenges. Those challenges include:
- Poor Data Quality. Inaccurate, inconsistent or incomplete data sets resulting in poor quality insight and reporting.
- Data Skills Shortage. Insufficient skills, limited capability and/or poor data literacy hindering business efficiency.
- Uninformed Decisions. Insufficient, inaccurate or unobtainable data to support decision making.
- Data Governance and Security. Adhering to governance, regulation and security requirements.
In security and policing there is an appetite for more data driven insights with decisions being more data- and evidence-driven across departments. The challenge is knowing where to start.
These departments are data rich (with probably too much data), with large volumes of data sets available. But policing end users are constrained by the transactional nature of these data sources, which inherently includes lead times on responses and not easily tailored to get the quality of data needed. With an underpinning theme of being reactive instead of proactive, and even sometimes, that data is not used to inform decision-making at all.
Cross-agency collaboration is also a key goal, as effective policing often requires information sharing among multiple agencies at the local, regional, and national levels. Overcoming jurisdictional barriers, establishing standardized protocols, and building trust among stakeholders are all essential for successful data sharing initiatives.
To fully harness their data, users strive to be able to get better insights from the data to make better decisions, have data relevant to their role and the ultimate nirvana of understanding the impact on policy.
Where are you on your data transformation journey?
With Generative AI hype near its peak, savvy organisations should try and avoid the frenzy. The no-regrets move is to put in place firm foundations for a long-term strategy.
Now is the time to review your organisation’s data maturity and readiness. What’s needed is a methodical assessment that moves through data maturity systematically:
- Envision – Having a clear mission, vision and strategy, aids in building the roadmap and the creation of an evidenced business case to support the investment in data harnessing.
- Enable – The definition and build of your data target operating model - platforms, data profiling & engineering, people, processes, and importantly a Common Data Model.
- Enhance - Now driving your business forward though the enrichment of data, at-scale delivery of cleansing, data quality and visualisation which will achieve increased business intelligence as the new normal for a data-enriched department.
- Evolve – Shifting the view from the past to the future, anticipating change rather than reacting to it, embracing AI tooling.
Common Data Model as the foundation
Central to the realisation of undertaking a data transformation journey is the adoption of a Common Data Model (CDM), which serves as a foundational framework for data integration, analysis and informed decision-making.
A CDM is a blueprint for data within an organisation or service area which brings together disparate data sources and acts as a guide for data usage within an organisation. It provides reliable, consistent and documented data standards, establishes “one source of truth” for improved reporting and richer insights, develops data literacy by giving users transparency in data and a single language.
People should be able to ask questions of their data and get answers that make sense and that they know they can trust. A CDM creates a long-term foundation for future projects and ambitions and is directly tied to tangible outcomes, making it easier to measure the value that is delivered.
It also allows for the easy sharing of data across departments, as data is clearly defined and structured. A CDM can help address several data challenges in security and policing by providing a standardized framework for integrating, analysing, and sharing data across different systems and agencies, where traditionally entities have operated in silos, leading to fragmented data landscapes that hinder comprehensive analysis and holistic decision-making.
For example, a CDM can standardise the format and structure of incident reports, enabling seamless integration of data from multiple sources such as law enforcement agencies, emergency services, and public safety departments. This allows for more accurate and comprehensive analysis of crime trends, hotspots, and patterns to inform resource allocation and proactive policing strategies.
Evolve and solve with AI
Now you have the fundamental foundations in place, your organisation can really assess how to evolve and focus on you will achieve most value in the adoption of AI tooling, particularly GenAI. According to research by Sopra Steria Next, the generative AI market is expected to increase tenfold by 2028 to around $100 billion, equating to annual growth of 65%.
The Sopra Steria Next recommended approach is to -
- Prioritise “no regrets” moves, by selecting, fine-tuning and deploying apps on existing foundation models on non-core use cases, leveraging potential new functionalities developed by your existing IT partners, thus focusing on basic task enhancement without or limited use of proprietary data.
- Put a few bets on “golden” high value use cases, by enhance existing GenAI models with proprietary data, developing long-term value-creating asset through use of proprietary data, and sourcing of relevant AI talent and skills.
- Think twice to qualify “diamonds” by assessing realistically the conditions for developing your in-house GenAI model by keeping complete control over model development and proprietary use, securing major IT infra/hardware investment.
The wrapper around all of this is implementing trustworthy and sovereign AI solutions. Responsible AI is needed to make it ethical, trustworthy and futureproof against regulatory requirements. This is particularly critical in security and policing, for example, to avoid algorithmic bias in historical data, leading to discriminatory outcomes. Ensuring fairness, transparency, and accountability in algorithmic decision-making is vital to avoid reinforcing existing disparities in law enforcement practices.
Conclusion
Data is the DNA of an organisation, so taking a taking a data first approach is key. By first unlocking the value of the data you have, it will provide allow better decisions. Establish your data strategy, tune your data management structure, onboard your people, upgrade your data technology foundations…all of which is underpinned by a Common Data Model as a foundation for interoperability, consistency, and collaboration across disparate data sources and stakeholders - ultimately improving the efficiency, effectiveness, trust and accountability of security and policing operations.
Then, determine the outcomes to be achieved and how data and AI can help, not forgetting the problems you’re trying to solve. Don’t get blinded by AI!
“The challenge lies in implementing this technology gradually, striking the right balance between proactivity and control, and coordinating the technological, regulatory and human dimensions involved. At a time when everyone is talking about ‘humans augmented by AI’, our experience with our customers in 2023 has convinced us that the key to success is to ensure that GenAI also remains ‘augmented by humans’.” (Fabrice Asvazadourian, Group Consulting CEO, Sopra Steria Next)