AI/Machine Learning Briefing Paper

Business Brief

Artificial intelligence (AI) is the general term referring to the development of computer systems to do “intelligent things”. By intelligent things, we refer to tasks that take some form of intellectual ability comparable to that of humans. This can be anything from voice and video recognition, data analysis, virtual personal assistants and much more. Artificial intelligence is largely based upon a numerous disciplines, such as philosophy, psychology and computer science. These disciplines are combined to produce new and enhance existing artificial intelligence based systems.

Machine learning however, is a subset of Artificial intelligence, with one specification. Machine learning aims at developing systems that perform tasks based off of experience. Essentially, machine learning entails programs to execute and forecast tasks that aren’t explicitly hard-coded into the system. Machine learning is already prevalent today across countless industries including, but not limited to, healthcare services, retail, financial services and search engines such as Google.

In addition, it is important that AI has many more branches other than machine learning. Some of the most prevalent are probabilistic AI, statistical AI, nouvelle AI and evolutionary computation. All of the branches (listed and not) contribute to the development of AI and will be critical in creating infallible tech.

Technical Brief

AI systems are powered by complex algorithms that try to perform tasks humans refer to as intelligent. These algorithms are integrated to function and execute according to the requests posed. The major tasks artificially intelligent machines are used for are calculations, information processing and automated reasoning.

Traditional AI machines perform based off rules and procedures hard-coded by the developers. AI machines take in their environment through various modes, such as auditory and visual, and perform tasks to achieve or help achieve a goal. Given a specific request, the task the machine performs is unambiguous to the user. To be specific, depending on a user’s request, they expect a certain result each time. A simple example of this would be if users request an action, A, they can expect a same result, X, each time.

Machine learning takes AI one step further by performing tasks which are derived from the machine’s experience and not hard-coded into the system. These algorithms, often categorized as supervised or unsupervised, generate data or responses that may not necessarily be foreseen by the user. Machine learning is formed on the basis that systems can identify patterns or learn from data to perform a function with little to no human mediation. A common example is Netflix providing suggested or relevant search options based off a user’s previously watched movies/shows.

Industry Use

Artificial intelligence is already being used within the Government of Canada (GC) across multiple departments. For example, the Communications Security Establishment uses AI to recognize forms of cyber threats. Once these threats are recognized, machine-learning algorithms adapt and evolve on their own to prevent such threats from compromising data. Other organizations, for instance, Veterans Affairs, CBSA and the National Research Council use their own forms of AI to help with tasks within the GC. Depending on the need, these tasks vary from examining video surveillance, analyzing data sets, providing services to clients and more.

Although there are many instances of AI within the GC, there is still a huge opportunity for other departments to introduce it and help enable a more efficient and productive GC. Among other things, a huge benefit to AI for financial committees is that it will be able to detect and prevent fraud at an incredibly fast rate. AI will also help departments produce models and analytics at a rate never seen before. In addition, AI will be able to support workers by providing aid in border security, healthcare, cyberwarfare, infrastructure planning, personal digital assistants and more. The computational capabilities will make data mining and retrieval much easier and powerful.

The emergence of artificial intelligence and all its subsets will provide the GC with support in almost all areas. AI will save the GC an abundance of resources which can be used for other priorities and projects. The advantages of AI will help the entire GC in delivering better services to its clients, stakeholders and all Canadians.

Canadian Government Use

Artificial intelligence is already being used within the Government of Canada (GC) across multiple departments. For example, the Communications Security Establishment uses AI to recognize forms of cyber threats. Once these threats are recognized, machine-learning algorithms adapt and evolve on their own to prevent such threats from compromising data. Other organizations, for instance, Veterans Affairs, CBSA and the National Research Council use their own forms of AI to help with tasks within the GC. Depending on the need, these tasks vary from examining video surveillance, analyzing data sets, providing services to clients and more.

Although there are many instances of AI within the GC, there is still a huge opportunity for other departments to introduce it and help enable a more efficient and productive GC. Among other things, a huge benefit to AI for financial committees is that it will be able to detect and prevent fraud at an incredibly fast rate. AI will also help departments produce models and analytics at a rate never seen before. In addition, AI will be able to support workers by providing aid in border security, healthcare, cyberwarfare, infrastructure planning, personal digital assistants and more. The computational capabilities will make data mining and retrieval much easier and powerful.

The emergence of artificial intelligence and all its subsets will provide the GC with support in almost all areas. AI will save the GC an abundance of resources which can be used for other priorities and projects. The advantages of AI will help the entire GC in delivering better services to its clients, stakeholders and all Canadians.

Implications for Departments

Shared Services Canada

Value proposition

Shared Services Canada can gain a lot from AI/machine learning. Committees dedicated to cyber security can use AI to integrate various independent alerts to detect incidents. In addition, automation and machine learning can be used to remediate and prevent cyber incidents without human intervention.

Following, all branches within SSC can use AI to retrieve and analyze data for their own specific needs. This will provide the public servants with support to produce precise and fast deliverables that will contribute to the overall production of SSC. The correct implementation and integration of AI will help SSC in the four areas of action known as Service, Secure, Manage and Community.

Challenges

The implementation of AI/machine learning technologies provides some challenges that will need to be kept in mind. The services AI provides need to be fully secured before operational. As AI is still relatively new, there can be many unforeseen incidents that can arise. SSC needs to ensure that security, equity law, values and ethics are all maintained to the highest degree.

In addition, machine-learning tech presents many hazards within itself. A machine that can think on its own may do so in un-predictable ways. This can result in a machine evolving on its own to generate erratic results which can be detrimental to SSC, the GC and Canadians.