Disclaimer: Please note that outputs generated from artificial intelligence models are not always accurate and can generate incorrect information, also known as hallucinations. We strongly recommend independently verifying AI-generated results to ensure accuracy and complying with your national or agency-specific AI guidelines for responsible use.
What is generative AI? How does generative AI work?
Generative Artificial Intelligence (GenAI) is a type of artificial intelligence that can create new content, such as text, images, or audio, by learning from existing datasets. Distinct from traditional AI, which makes predictions or decisions based on past data, GenAI goes further by producing new content that mirrors the patterns and structures learned from its training data. The technology behind GenAI allows the system to understand and create text, images, and sounds that fit the context the user provides, producing responses that feel natural and human-generated. These GenAI tools can gather and synthesize information in ways that mimic how we conduct research, analyze documents, or aggregate information. Simply enter your instruction for what you want (i.e., provide a prompt) into the GenAI tool and it will create a tailored response. For example, if you’re using a chatbot, you could ask it to summarize a policy or draft an email for you (e.g., “Summarize key points of the Paris Agreement”).
This technology is reshaping productivity in the workforce. When used properly GenAI could accelerate certain tasks, sometimes by a large degree. It also introduces new challenges and risks.
Government agencies worldwide have an opportunity to modernize the public sector through the adoption of GenAI. By incorporating GenAI into everyday government work can help civil servants in reducing manual tasks, speed up data processing and analysis, automating document generation, to name a few. For civil servants, the advantages of using GenAI tools extend to even labor-intensive jobs. For instance, the U.S. Federal Highway Administration (FWHA) is developing AI that would automatically inspect transportation infrastructure, providing real-time information for city planners and maintenance crews (FWHA, 2024).
Common Acronyms and Definitions
AI
Artificial Intelligence is a broad group of technologies that focus on developing machines to perform and solve tasks that resemble human behavior and intelligence.
ML
Machine Learning is a method where computers learn and perform tasks by finding patterns in data with little to no human supervision, unlike typical computer systems that rely on specific programming for each task. Through different techniques and methods, machine learning trains systems and models to make decisions or predictions, mimicking aspects of human reasoning. It works similar to how a public transportation system might improve its routes over time, learning from past passenger behavior and identifying patterns in travel time.
AI Models
An AI model is a program that has been trained to perform a specific task. They leverage patterns in their training data to identify similar features in newly presented data to form responses or specific outputs.
Algorithms
Within Artificial Intelligence, algorithms are a set of instructions that help systems learn on its own to make decisions and improve over time. They operate within AI models to recognize patterns based on the collection of data inputted, similar to guidelines for the model to understand what to do with all of the data given.
LLMs
Large Language Models are a type of AI model that can handle massive amounts of text to understand language patterns. These models are trained on large datasets to generate contextual language to perform tasks like answering questions, summarizing text, and engaging in a conversation.
Prompt
A prompt is the information you, as a user, give to the AI model, like asking a colleague to complete an assignment. It can be a specific ask or question you instruct the AI model to perform by entering text or speaking to the system.

Available Generative AI Tools
Disclaimer: Generative AI tools are constantly evolving, and the field is rapidly producing new and improved tools. While the tools mentioned are popular now, they might change or be replaced as the technology evolves.
Currently, it is more effective for users to rely on a mix of AI tools tailored to specific tasks, rather than depending on a single application. This approach ensures that each task is supported by the best-suited tool.
Here’s a breakdown of some of the most popular GenAI tools you can use today and how they work:
Text Generation Tools
You’ve likely used text generation tools without even realizing it, whether it was chatting with a customer service assistant on a shopping site or using the help chat box on the IRS website. These chatbots, which interact with users through human-like conversations, are often powered by Large Language Models.
A few commonly used tools today are in the table below:
Tool | Key Features | How to Use It |
---|---|---|
Copilot Microsoft | Integrated within Microsoft 365 apps like Excel, Word, and PowerPoint to securely draft, summarize, or answer questions that are related to their content in the app. | Activate Copilot in a Microsoft 365 app and type an instruction that relates to the content within the document. |
ChatGPT OpenAI | Can answer questions, brainstorm ideas, draft content | Ask questions or give instructions to request tasks like summarizing documents directly on the website |
Gemini | Uses Google search engine and can create text and images | Use Gemini in Chrome by typing @gemini in the search bar, or in the web app (gemini.google.com), and connect your google workspace account |
Claude Anthropic | Analyzes and summarizes long documents while answering questions and assisting with writing tasks | Upload or paste documents, then ask questions or request a summary |
DeepL Translate DeepL | Translates text across over 30 languages with a focus on contextually accurate translations | Paste text or entire documents to translate the content into a different language |
Many text generational tools are becoming increasingly integrated into existing products and services, making them more accessible to civil servants.
Image, Video, and Audio Generation Tools
These tools simplify the process of creating images, audio, or video based on the descriptions you provide. They can be used to make images to support program materials, assist with video editing for training modules or outreach campaigns, or produce auditory translations.
A few commonly used tools today are in the table below:
Tool | Key Features | How to Use It |
---|---|---|
Midjourney Midjourney, Inc. | Creates artistic and detailed images from text descriptions. | Type a description to generate images |
DALL-E OpenAI | Generates images from text and can edit existing images | Enter a detailed prompt or upload images for editing or expanding content |
Runway Runway AI, Inc. | Provides AI tools for video editing, image creation, background removal and visual effects. | Use prebuilt tools or upload videos/images to apply effects and make edits |
Eleven Labs Eleven Labs | Generates realistic AI voices from text. Specializes in voiceovers, narration, and voice cloning | Type or paste your text, select a voice, and generate audio for voiceovers or narration |
KitsAI KitsAI | Focuses on generating music, including creating original compositions, sound effects | Enter prompts for the type of music or sound you need |
With the myriad of GenAI tools released each year, choosing the one that best fits you and your department requires an in-depth evaluation in comparing different GenAI tools. What works for one agency or team may not be suitable for another due to differing work functions and varying regulatory requirements across agencies.
Here are some key questions and considerations to help guide your search:
While civil servants can use publicly available GenAI platforms, the growing adoption of GenAI has sparked significant international investment in custom-built AI models, such as India’s budget approval of Rs 10,371.92 crore ($1.2 billion) for the IndiaAI Mission initiative (PIB, 2023).
With global spending in AI reaching $154 billion in 2023, the public sector is in the early stages of a major shift as governments are investing in GenAI solutions that protect and maintain data privacy, security, and compliance to meet government operational needs.
Sources
-
Association of Southeast Asian Nations (ASEAN). (2024, February). ASEAN guide on AI governance and ethics. Retrieved from https://asean.org/wp-content/uploads/2024/02/ASEAN-Guide-on-AI-Governance-and-Ethics_beautified_201223_v2.pdf
-
U.S. General Services Administration, Centers of Excellence. (n.d.). AI guide for government: What is AI? Key terminology. Retrieved from https://coe.gsa.gov/coe/ai-guide-for-government/what-is-ai-key-terminology/
-
Central Digital and Data Office. (n.d.). Generative AI framework for HM government. Retrieved from https://assets.publishing.service.gov.uk/media/65c3b5d628a4a00012d2ba5c/6.8558_CO_Generative_AI_Framework_Report_v7_WEB.pdf
-
Boston Consulting Group. (2024, March 26). Generative AI for the public sector: The journey to scale. Retrieved from https://www.bcg.com/publications/2024/gen-ai-journey-to-scale-in-government
-
California Government Operations Agency. (2023, November). Benefits and Risks of Generative Artificial Intelligence Report. Retrieved from https://www.govops.ca.gov/wp-content/uploads/sites/11/2023/11/GenAI-EO-1-Report_FINAL.pdf
-
Press Information Bureau. (2023, December 3). Cabinet Approves Over Rs 10,300 Crore for IndiaAI Mission, will Empower AI Startups and Expand Compute Infrastructure Access. Retrieved from https://pib.gov.in/PressReleasePage.aspx?PRID=2012375#:~:text=Funds%20approved%20by%20the%20Cabinet,across%20all%20strata%20of%20society.
-
U.S. Department of Transportation, Federal Highway Administration. (2024). Employing Artificial Intelligence (AI) to Enhance Infrastructure Inspections. Retrieved from https://highways.dot.gov/research/publications/EAR/FHWA-HRT-24-055