Vanguard Magazine

the lASt woRD 46 JUNE/JULY 2018 www.vanguardcanada.com nology – algorithms and code – that rep- licates 'human' thinking by utilizing digi- tized human behaviour heuristics, learning from the outcomes and creating 'intelli- gence' for integration, thereby improving and refining the process. It's all based on patterns, which coincides well with the processes we observe in policing. Contrasting with 'natural learning' – how humans learn based on their neurolo- gy, innate behaviours, predispositions and environmental experiences that become learning experiences – AI uses technology to associate actions to desired outcomes and retention (memory) and learning to help dictate future actions. Here, it's im- portant to differentiate AI from forecast- ing systems; AI is generative, resulting in predictive, adaptive and more efficient responses that can be extended to other problems that need to be solved, making AI inherently useful to policing. AI and policing's problem with data When applied through technology, AI can manifest in many functional aspects of data management, improving data analyt- ics and management for policing. Some of the most anticipated are: • Crime prevention and interdiction through predictive policing • Investigative functions in analysis, search and correlation • Associative functions for data between programs, case management and special- ized units • Intelligence gathering, from tip consoli- dation to very complex field intelligence All data types, once characterised and catalogued, are enhanced in value and usability, whether GPS data, license plates, arrest records or weapons viola- tions or more sophisticated formats like detailed intelligence or field data, sur- veillance and body-worn camera files (audio/video) and biometric, forensic and material evidence. Whether the data is individually identifying or statistically informing of a group or area, related to intelligence, evidence or heuristic, the more integrated it is to other data, the more informative it is. Being able to integrate and correlate data is one step toward improvement, but AI's ability to rapidly and accurately take correlated data and generate associations and conclusions are where substantial ef- ficiencies are realized. In simple terms, when these data types are accessed and analysed by AI technologies, they can gen- erate associations that provide predictive conclusions on individual identities, of- fender risk profiles, and probable or preva- lent scenarios. With all its promise, there are limitations or drawbacks with AI, as with any tech- nology. One is that AI reaches conclusions that are often ambiguous, so data schema design at the front end will found the pri- mary integrity of results. Further, regard- less of its potential value, AI conclusions are only as good as the input data and its integrity – it is truly a matter of "garbage in, garbage out" as AI will not magically improve the quality of input data. Beyond these, the consideration of ethics, governance and legislative as other bastions of compliance and the acceptance of ambi- guity, as well as autonomy and errors (AI is not flawless), are also critical to planning and implementation prior to embarking on transformations that include AI. What is blockchain and why it might be important? As soon as we hear the term 'Blockchain', we think of crypto-currencies, like Bit- Coin. But blockchain is actually a design concept that, by using other structures, assures the integrity, control and unique- ness of data: the same conditions required in evidence management and chain of cus- tody, especially where evidence requires transfer or custodial access. Providing a means to create open narratives and elabo- ration on data stores, blockchain is a bit of a dark horse to many sectors, but its defin- ing controls for data integrity are becom- ing clearer outside crypto-currency. Blockchain at its base is essentially a dis- tributed database of records that contains all designated transactions performed or shared among specific users. Every trans- action is a verifiable record of every single transaction made, making it useful in not only record keeping, but chain of custody, forensic identifiers, ID systems and cloud storage. As AI and blockchain progress in their maturity and adoption, policing will see a lot more them and in more creative uses not yet analyzed. New technologies that allow for open and portable identity, au- tonomous agents, 'always on' devices and more robust networking architectures will create unique challenges in crime preven- tion, detection and interdiction for police strategic governance in the form of privacy regulations and legislation. As the landscape of cyber and e-crimes change and their impacts on traditional crime types become clearer, both AI and blockchain's functional concepts stand to improve policing through intelligent and adaptive thinking, associating and learning, if thoroughly planned and implemented. Valarie Findlay is a research fellow for the Police Foundation (USA) and has two de- cades of senior expertise in cybersecurity and policing initiatives. She holds a Masters in Terrorism Studies from the University of St. Andrews, and her dissertation, "The Impact of Terrorism on the Transformation of Law Enforcement" examined the trans- formation of law enforcement in Western Nations.

• Cover