Learn the basics of artificial intelligence

“AI” refers to a group of technologies that can perform complex tasks normally requiring human intelligence, such as learning behaviours, making sense of spoken language or solving problems. AI first emerged as a field of computer science in the 1950s, but it has recently become far more capable and widely used because:

• More data is available to train AI models, which improves their performance
• There’s access to better software, including open-source tools
• More computing power is available, at lower cost
• New models, algorithms and techniques have been developed
• AI tools have become highly marketed and mainstream

Sometimes it’s obvious when you use AI, like when you interact with applications like OpenAI’s ChatGPT. But AI is also found in many programs we use every day. For example, Microsoft Word uses AI when suggesting the next word in a sentence.

The B.C. government is currently using AI to provide better, faster and more efficient services for people living in the province. For example, the Ministry of Attorney General uses an AI Reviewer to validate, extract and process PDF information from Court Service Online forms directly into their database. If the tool is uncertain about any information on the form, it flags the form for human review and the input is completed manually. This tool has processed over 30,000 forms to date and frees up public service employees for more valuable work.

Common terms

• Algorithm: A set of rules or instructions used by machines and computers to achieve a specific goal
• AI system: A machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment
• Bots: Software applications which perform automated tasks
• Data: Information in structured or unstructured formats like text, numbers or images. AI systems get their value from processing massive amounts of data
• Machine Learning: Systems that learn from data and improve their performance by adjusting their approach over time, based on predefined success criteria. The models can learn and adapt without explicit instructions and are widely used in AI systems to improve performance
• Neural Networks: Algorithms inspired by the structure of the brain. Neural networks are used to find patterns and learn from data and can improve performance without being instructed
• Deep Learning: A subset of machine learning, essentially a neural network with 3 or more layers
• Generative AI (gen AI): A form of AI that generates new content (such as text, images or audio) based on natural language inputs called prompts. The content is what the model thinks is statistically likely to be the best response
• Large Language Model (LLM): LLMs power gen AI and other AI systems. They use machine learning to process vast amounts of data and generate human-like textual responses
• Prompt: The information, instructions or questions put into a gen AI tool to obtain specific results
• Hallucination: An AI-generated response containing false or misleading information, presented as fact

How information flows in an AI system

Branches and techniques

There is no single AI technology. A range of different techniques and approaches are used to solve different problems. New AI tools and uses will emerge as people and organizations invest more time and money into development.

Computer vision

Systems which make sense of digital images. Includes image recognition and machine vision.

Natural Language Processing (NLP)

Tools to understand and generate human language. They can analyze text or let you interact with software like a conversation, as seen in chatbots or gen AI.

• Learn about Chatbot Prompt Essentials (Government of New South Wales)

Robotic process automation

Tools automating repetitive tasks in a digital workflow such as inputting data, digitizing forms or connecting data sources.

AI assistants

Software enhancing productivity by streamlining tasks, automating workflows, connecting software, managing calendars, aiding decision-making, generating content and more.

Using AI responsibly

As AI becomes more advanced and widely used, there is a greater risk that it may perpetuate inequality or be misused, even unintentionally.

The public sector has a broad scope and reach when serving the public, and we are obliged to uphold high standards of conduct by using AI responsibly.

Before using AI tools, you should be familiar with:

• Draft guiding principles for the responsible use of artificial intelligence
• Appropriate Use Policy including the Application and Software Guide
• Interim guidance on generative AI

Considerations for responsible use

Bias and fairness

Outcomes from AI systems reflect the data used to build, train and refine the models they use. If the models have flaws or are trained on biased data, it can lead to unfair decisions or produce biased outcomes. This risk is especially acute with historical data which might be influenced by systemic racism and discrimination.

The BC Public Service needs to be aware of biases in the data used to develop AI models and tools. This includes monitoring and testing to make sure these models and tools operate without unfairly discriminating against groups of people or feeding existing inequalities.

Transparency and accountability

Organizations leveraging AI systems must take responsibility and accountability for actions, outputs and decisions made by those systems. They should track and maintain an inventory of what AI systems they are using and how.

We must maintain accountability, be transparent, and be able to explain how AI is used in the BC Public Service.

Privacy, security and governance

Any processing of personal data or confidential information needs to be governed and protected. You need to be mindful of what data you are putting into AI systems and where that data goes.

For example, AI tools are often cloud-based or use externally created resources, code or models, and the information you put into publicly available tools is often collected and used by the company creating the tool. Do not put confidential or personal information into a tool unless you are sure it’s secure.

Data sources and copyright

You must be mindful of the provenance (or source) of your data. Data provenance concerns the origins, ownership, collection and reliability of source data.

Be aware that AI datasets often contain vast and opaque sources, including personal information or copyrighted works, and that many tools source data from the wider internet which may not be directly applicable to B.C.

Accuracy and human oversight

Remember that AI systems aren’t infallible. Gen AI may create content that includes incomplete or false information, sometimes called hallucinations. Always have oversight from people involved with development and operation of AI systems, and fact-check the outputs.

Future opportunities and considerations

Opportunities

AI is being increasingly adopted by the BC Public Service and other governments around the world to enhance service delivery.

Improving data analysis

AI can provide new methods and faster analysis of complex data and information.

Delivering more efficient services and operations

Systems can enable automation of routine tasks, faster delivery of information and analysis, better accessibility and translation and more accurate forecasting of service demand.

Developing policy faster and better

Developments will allow faster and better analysis of public input, and increased business and stakeholder intelligence.

Enabling a more risk-based approach to regulation and compliance

AI can improve compliance monitoring and enable more targeted testing and inspections.

Considerations

Adoption of AI systems is increasing among all people and private companies. As AI systems become more common, the BC Public Service will have to navigate this changing environment carefully.

Private sector and timely adoption

We need to understand what tools are being used and what capabilities AI technology has in the private sector, particularly in operations and regulatory roles. It will take time for the BC Public Service to develop, procure or adopt AI solutions and approaches, so we need to be careful not to fall behind the private sector.

Ensuring data security

Protecting and managing our data will become even more important as data becomes more valuable and useful for AI tools, the amount of data increases and datasets become more linked together.

Maintaining public trust

AI will become more common in the public service, and there will be increasing expectations to leverage AI for delivering better services to the public and creating efficiencies. While we meet this challenge, we will need to assure the public that we are using AI responsibly and maintain their trust.

Delivering value

AI tools can deliver great value but won’t always be the best solution to a business problem. Make sure you consider the benefits, costs and risks of the AI tool and other solutions to pick the best way forward.

Countering manipulation and deception

AI can be used to generate and share misleading or false information. External entities may create or spread disinformation, misinformation or deepfakes using AI.

Managing new cybersecurity threats

AI technologies are becoming more sophisticated and can be used maliciously to create hard-to-detect threats. Bad actors can use AI to commit scams and fraud, or to attack our security systems. We need to be aware of this evolving landscape and act accordingly to protect ourselves and the public.

---