Skip to main content
All CollectionsAI Grant Writing Course
Module 2: AI for Grant Writing
Module 2: AI for Grant Writing

This module covers the application of generative AI tools to your grant-seeking work

Philip Deng avatar
Written by Philip Deng
Updated over a week ago

Introduction

In this module, we will explore the use of generative AI in grant-seeking and its potential benefits and drawbacks. We will delve into the different kinds of generative AI technologies that can be applied to the grant-seeking process, and how they can be used for more efficient and effective grant-seeking.

Throughout this module, we will present various strategies and techniques for responsibly incorporating generative AI into your organization's grant-seeking process. We will discuss the limitations of generative AI and the areas where human expertise and oversight is still necessary.

By the end of this module, you will not only understand the potential impact of generative AI on the grant-seeking landscape, but also be able to identify potential future developments in generative AI that could impact grant-seeking.


Learning Objectives

Assess the pros and cons of AI for grant writing

Describe the benefits and drawbacks of using generative AI in grant-seeking.

Compare and contrast different AI grant writing tools

Appraise different generative AI technologies and their potential applications in grant-seeking.

Strategize for responsible AI use

Develop a strategy for responsibly incorporating generative AI into your organization's grant-seeking process.

Recognize the limits of AI tools in grant-seeking

Understand the limitations of generative AI in grant-seeking and identify areas where human expertise is still necessary.

Envision the future of AI and philanthropic work

Explain the potential impact of generative AI on the grant-seeking landscape and identify potential future developments in generative AI that could impact grant-seeking.


Terminology

Use this section like flashcards

AI hallucination

When generative AI systems produce outputs that are illogical, nonsensical, or even offensive due to the limitations of the software

AI safety and responsibility frameworks

Guidelines and principles for the ethical and responsible development and use of AI technology

AI tools

Software applications that use AI technology to assist with various tasks, such as generating content or analyzing data

Generative AI

AI technology that is capable of generating text, images, or other digital content

Grant-seeking

The process of searching for and applying to grant opportunities to secure funding for an organization or project

Large language models (LLMs)

AI models trained on large datasets of text to learn patterns and relationships between words and phrases

Oversight

Supervision or careful management of a process or situation

Scalability

The ability of a tool to handle increased performance demands to meet the needs of an organization as it grows


Advantages and disadvantages of AI grant writing

The use of generative AI in grant writing can offer a wide range of benefits, including increased efficiency and the ability to generate a larger volume of grant proposals. By automating certain aspects of the process, such as generating the initial draft of a proposal or identifying potential funding sources, grant seekers can save valuable time and resources that can be allocated elsewhere.

However, there are also potential drawbacks to using AI in this process. One potential risk is that the proposals generated by AI may be too formulaic or lacking in creativity, which could decrease the chances of securing funding. Without human input, AI-generated proposals may not be able to capture the nuances of the grant seeker's organization or project, which could make the proposal less compelling to funders. To overcome these risks, it is extremely important to choose the right AI tools and to exercise careful oversight when deploying them.

How can AI grant writing increase efficiency?

When an organization is ready to apply for grant funding, and they have found a high-quality opportunity to pursue, the work of producing a grant application can begin. The most efficient organizations build upon existing work by repurposing language from recent proposals to produce new grant applications more quickly. Generative AI tools can supercharge this process by leveraging the incredible abilities of large language models to help locate and reformat grant proposal content in fractions of a second.

Locating content

Manual scenario

You're working on a grant application and you come across a question that you know you've answered before. The problem is that you can't recall which of the dozens of recent grant proposals in your content library contains the previous answer. You proceed to check file after file, searching through shared drives and inboxes looking for keywords and clues to help you locate what you’re looking for. Finally, after a few minutes, or in the worst cases, several hours or days, you locate the passage, which you copy and paste into your new draft.

AI scenario

If you are working on a grant application and need to find a previously answered question, consider using Grantable, a generative AI grant writing software. Grantable can instantly search through all the content in your library and retrieve meaningful excerpts that you may wish to reuse. This saves you from having to go through each document and folder individually. You can search your entire library at once and either select excerpts yourself or allow the AI to recommend resources. The entire process usually takes only a few seconds.

Reformatting content

Manual scenario

When crafting a grant proposal, it's common to reuse text from previous proposals. However, this can result in excess material that needs to be trimmed, updated, and adjusted to fit the new application question. Meeting vastly different length limits imposed by different application formats can be a significant challenge. For example, your source material may come from a proposal with a much longer format, requiring considerable effort to condense the text while preserving important information to meet a shorter word limit. The time required for this process can range from a few minutes to many hours, depending on the situation.

AI scenario

Grantable, a generative AI grant writing software, and other applications like ChatGPT excel at taking source material you provide and quickly reformatting it to address the various points in a grant application question and meet specific formatting requirements. These AI can also help revise and reformat the text to change details, perspective, tone, and more. Each of these processes takes only seconds, giving grant professionals more time and opportunities to curate and refine the overall proposal.

The bottom line

According to the 2023 State of Grantseeking Report produced by GrantStation, the top challenge reported by grant-seekers was insufficient time and capacity to pursue grant funding. Generative AI tools like Grantable can significantly reduce the time and energy required to produce high-quality grant applications, enabling organizations to apply for all the opportunities they may qualify for.

While there is no way to guarantee a successful funding outcome, grant-ready organizations that pursue well-aligned grant opportunities and submit as many qualified applications as possible are more likely to see success over time. By using generative AI tools like Grantable to store grant proposal content in a centralized and intelligent library, and leveraging generative AI writing features to reduce the amount of staff capacity per application, organizations can increase their chances of successful award outcomes.

Manual

AI

Manual document search

Instantly search and retrieve excerpts

Time-consuming

Takes only a few seconds

Need correct keywords

Locates content based on context and meaning

Copy and paste

Select excerpts and synthesize

Constant reformatting

Revise and reformat in seconds


Comparing different AI writing tools

In recent years, generative AI has gained more attention with the release of various tools like GPT-3 and ChatGPT. These tools have made it easier for people to experience the power of generative AI firsthand. The use of generative AI in grant writing is a relatively new field of exploration, and there are a few different tools available that are specifically designed for this purpose. In this section, we will compare some of the most popular AI grant writing tools and discuss their relative strengths and weaknesses.

ChatGPT and Claude

ChatGPT and Claude are free chatbots, which make them accessible to all kinds of organizations. When you use these bots, you will need to create an account on the platform's website. The sign-up process typically involves providing basic information such as name, email address, and password. Once you have created an account, you can log in and begin using the tool.

To create content, you will need to provide a prompt or starting point for the AI to work with. This could be a question, a topic, or a general idea. The chatbots will then generate a response based on the prompt, which you can review and edit as needed. The AI can also provide suggestions for related content or additional ideas to help you expand on your initial prompt.

Overall, the user experience of these chatbots is designed to be simple and intuitive, with a user-friendly interface and helpful features to guide you through the content creation process.

ChatGPT

To use a free AI chatbot as a grant writing tool, follow these steps:

  1. Create a free account on the ChatGPT or Claude website.

  2. We recommend reading the privacy policy so you understand how your data is used.

  3. Create a new chat thread

  4. Begin by prompting the chatbot to act as a grant writer

  5. Example prompt: “Act as a helpful grant writing assistant”

  6. Next, provide the chatbot with source material you’d like it to use to help you draft grant proposal content

  7. Example prompt: “The following is grant proposal content I’d like you to use for the purposes of helping me: [PASTE CONTENT]

  8. Tip: Keep length limits in mind and share specific excerpts of your material, rather than large un-curated selections

  9. Finally, prompt the bot to draft content for you

  10. Example prompt: “Using the content I’ve provided, please draft a response to the following grant application question: [PASTE GRANT APPLICATION QUESTION]

  11. Optional: Once the chatbot has drafted a response, you have the option of asking it to revise and reformat the text

  12. Example prompt: “Please revise your last answer to be 100 words or less, and use a more conversational tone.”

  13. Many people will use chatbots to bring the grant narrative to a certain level of completion at which point they will copy and paste the text to another word processor to finalize

To use chatbots as a strategic tool, follow these steps:

Repeat steps 1 and 2 from above

  1. Prompt the bot to offer strategic advice to give you ideas of how best to construct your grant application

  2. Example prompt: “Please share 10 examples of ways food banks play a vital role in the community”

  3. Tip: Use the outputs to help you generate ideas for how best to convey the impact of the organization’s work

  4. Example prompt: “Please suggest ways to respond to the following grant application prompt strategically, thoroughly, and in a compelling way: [PASTE GRANT PROPOSAL QUESTION]

  5. Tip: Keep the outputs nearby as you work, to help you stay on track

  6. Tip: Consider feeding these recommendations back into ChatGPT when requesting the chatbot’s help to write

Conclusion

Overall, chatbots are an incredibly powerful tool that can be a useful addition to anyone's toolkit. They are free, easy to use on both desktop and mobile devices, and extremely flexible in terms of its capabilities.

The biggest drawback of using a chatbot for grant-seeking is the labor-intensive and repetitive process of providing context to the model before asking for outputs. These bots are intended to be general-purpose interfaces to assist hundreds of millions of people worldwide with a wide range of queries and is therefore not optimized for understanding a specific organization or workflow.

Feeding it context can significantly improve the focus and outputs of the tool, but it may still require searching through multiple files and folders to find the right content. These chatbots store your interactions as chat threads, similar to text message conversations, which are excellent for managing conversations but inefficient for storing document information, such as grant proposals.

As a chatbot, the interface is designed for people to engage with the bot in back-and-forth dialogue. It is not structured for writing long-form documents, like grant proposals. There is no ability to edit or format like in a word processor, and it can be difficult to review a grant proposal when it has been rendered in the format of a conversation, rather than as a document with sections, prompts, and responses.

Grant writing with AI chatbots pros and cons

Pro

Con

Powerful and flexible tool

Requires labor-intensive and repetitive process of providing context

Free and easy to use on desktop and mobile

Not optimized for understanding a specific organization or workflow

Can be a useful addition to anyone's toolkit

Difficult to locate source material

Excellent conversational format

Inefficient for storing document information, such as grant proposals

Can generate helpful outputs and suggestions

Not structured for writing long-form documents, like grant proposals


Grantable

Grantable is a generative AI software designed specifically for grant-seeking. Similar to leading AI chatbots, it utilizes the latest large language models to provide comprehension and writing abilities. Additionally, it comes with a smart content library and a user-friendly word processing interface that simplifies grant content production.

Grantable word processor with AI

To use Grantable, you must create an account using a valid email address. Many people start with a free account and upgrade later if they find the tool useful. Users must upload writing samples to their account, preferably a recently completed grant proposal, which Grantable processes as source material.

To begin working on a new grant proposal, create a new file in Grantable as you would in a typical word processor. Grantable's AI assistant can draft content at any point in the document and provide suggested source material snippets to respond to prompts. You can edit anything on the page, including the outputs of the Grantable assistant, and ask for revisions as well.

Grantable continuously updates all the content uploaded and created in an account, enabling it to locate the best source materials for a given task. With more content created, the system becomes better at drafting responses.

To use Grantable as a grant writing tool, follow these steps:

  1. Create an account on the Grantable website using a valid email address

  2. Grantable offers a free trial, and users can upgrade to a paid plan to continue using the tool after the trial runs out

  3. Upload writing samples to your account, preferably a recently completed grant proposal, which Grantable processes as source material

  4. Create a new file in Grantable as you would in a typical word processor

  5. To prompt Grantable to generate content, select a text area in the document and engage the AI assistant

  6. Locate and designate source materials from your library for the AI to use, or have the AI suggest source materials

  7. Grantable's AI assistant will generate a response based on the prompt and source materials, which you can review and edit as needed.

  8. Edit anything on the page, including the outputs of the Grantable assistant, and ask for revisions as well.

  9. Once you have completed your grant proposal, you can export it as a .docx document for additional formatting or submission.

  10. Grantable continuously updates all the content uploaded and created in an account, enabling it to locate the best source materials for a given task. With more content created, the system becomes better at drafting responses.

Conclusion

While both Grantable and chatbots are generative AI tools that can be used for grant writing, there are significant differences between the two.

One of the biggest advantages of Grantable is that it is specifically designed for grant-seeking, while AI chatbots are a general-purpose AI tools. This means that Grantable is optimized for particular work and information cycles in the grant-seeking process. As a result, Grantable is more likely to generate high-quality, coherent grant proposals that are tailored to the needs of the specific grant opportunity.

Another advantage of Grantable is its smart content library. Unlike other chatbots, which require users to search through multiple files and folders to find the right content, Grantable continuously updates all the content uploaded and created in an account, enabling it to locate the best source materials for a given task. With more content created, the system becomes better at drafting responses.

Finally, Grantable provides a user-friendly word processing interface that simplifies grant content production. This makes it easier for users to manage grant proposals and make edits to their content as needed, something that is not possible with with chatbots at this time.

While chatbots can be a useful addition to anyone's toolkit, its general-purpose nature means that it may not be the best choice for grant writing.

Chatbots

Grantable

User-friendly interface

Yes

Yes

Can generate helpful outputs and suggestions

Yes

Yes

Can draft high quality written outputs

Yes

Yes

Optimized for grant-seeking

No

Yes

Smart content library

No

Yes

Ability to locate and reformat grant proposal content

No

Yes

Ability to edit and format writing

No

Yes


Adding AI to your organization

If you're considering incorporating generative AI into your grant-seeking process, there are several important factors to consider. Here are some key steps to take to ensure that you are making the best use of generative AI in your organization:

  1. Identify the areas of your grant-seeking process where generative AI can be most helpful, such as content creation and review, proposal formatting, and content management.

  2. Research the different generative AI technologies available and their potential applications in grant-seeking.

  3. Choose the right generative AI tool for your organization's needs and budget.

  4. Train your team on how to use the generative AI tool effectively and responsibly. It's important to ensure that everyone understands the limitations of the tool and that it should be used as a supplement to, not a replacement for, human expertise.

Identify the areas of your grant-seeking process where AI can help

Brainstorming

  • Problem: Everyone experiences writer's block at some point. This can be especially difficult for grant-seekers who are just starting out and struggling to know what to write, or for highly experienced individuals who are fatigued or lacking inspiration.

  • Solution: Generative AI tools can be useful for grant-seekers in brainstorming and strategic planning. By providing prompts or starting points, these tools can generate ideas and suggestions that can help guide the grant-seeker's thought process and provide new perspectives on the problem at hand. For example, ChatGPT can provide strategic advice to give grant-seekers ideas of how best to construct a response in grant applications.

Composition

  • Problem: Good writing generally requires a substantial amount of time to produce. Even if you know the material, it can take a while to get the words onto the page, and more time to edit them into final form.

  • Solution: Generative AI can be used for content creation in grant proposal writing. By automating certain aspects of the process, such as generating the initial draft of a proposal, grant-seekers can save valuable time and resources that can be allocated elsewhere.

Formatting

  • Problem: Every funder has different application formats and requirements, even if they’re requesting the same information. It can be a huge time-waster and hassle to reformat your content to meet these different situations.

  • Solution: Generative AI can help format grant proposals to meet different application requirements. It can recognize specific formatting requirements, such as tone and length limits, and adjust the proposal accordingly. This can save time and ensure that the proposal meets the specific needs of the funding opportunity.

Content management

  • Problem: Organizations and the tools we use are constantly evolving. Over time, written content tends to end up strewn all over different laptops, inboxes, and shared drives. This makes finding a particular document, page, paragraph, or sentence a time-consuming activity.

  • Solution: Generative AI systems that come equipped with a smart content library, like Grantable, leverage the power of large language models to search across vast data sets. This means that you don’t need to do as much organizing to keep track of everything.

Program/Client management

  • Problem: For large organizations with many tracks of grant-seeking or professional grants consultants who work with many organizations to apply for grants, keeping content organized and separate across all of these workspaces can be problematic and quickly get out of hand.

  • Solution: Generative AI systems like Grantable allow you to create workspaces that separate content, which are easily navigable.

Researching different AI tools

Keeping in mind the needs identified above (brainstorming, composition, formatting, content management and program/client management), consider using the follow resources periodically to learn about the latest AI tools for each situation.

Search engines

Use your the search engine of your choice (Google, Bing, etc.) to find tools to try by searching terms such as “generative AI tool for [INSERT YOUR NEED]”.

This website continually aggregates new AI tools and categorizes them by use. Search for the kind of tool or help you need and review the results.

Study the websites you find and consider important factors such as:

  • Functionality: Does the tool provide the functionality you need to achieve your goals? Are there any key features missing that would be critical to your organization?

  • Ease of use: Is the tool intuitive and user-friendly? Will your team be able to use it without requiring extensive training?

  • Integration: Can the tool be easily integrated with other tools and systems that your organization uses?

  • Customization: Does the tool allow for customization to meet your organization's unique needs?

  • Scalability: Can the tool scale to meet your organization's needs as it grows?

  • Security: Does the tool have appropriate security measures in place to protect sensitive data?

  • Support: Does the tool provide adequate support resources, such as documentation and customer service, to ensure that your team can use it effectively?

  • Cost: Is the tool affordable for your organization, and does it provide good value for the cost?

Follow technology journalism

We recommend the following podcasts as fun ways to stay up to date on the latest and most important discussions and trends in technology. This is a great way to hear about new tools and hone your ability to critique them.


How to use AI responsibly

When working with generative AI tools like Grantable, it's important to use them responsibly and collaboratively. Ensure that everyone on your team understands the limitations of the tool and how it should be used as a supplement to, not a replacement for, human expertise.

Grantable has endorsed the Framework toward Responsible AI for Fundraising developed by Fundraising.AI and we believe these guidelines can work well for grant professionals learning to add AI tools to their work.

The Fundraising.AI collaborative is a member-driven initiative supporting those working within the fundraising profession with the opportunity to collectively learn about Responsible AI, demonstrate their leadership around the subject, support best practices of Responsible AI applications, and support building a thriving charitable giving sector.  The Framework for Responsible AI for Fundraising is intended to maximize the benefits of AI for fundraising purposes while minimizing the risk of damage to the hard-fought public trust of the nonprofit sector.

There are many different AI safety and responsibility frameworks being proposed, and we’ve chosen to share this one because it is specifically adapted for fundraising. Here is the framework:

Privacy and Security

Fundraising AI Actors must protect personal and sensitive data by following robust security standards within our respective roles, maintaining compliance with relevant data protection regulations, and respecting the privacy of donors, beneficiaries, and stakeholders.

These principles should be part of all phases of the AI system lifecycle, including;

  1. Consent,

  2. Control over the use of data,

  3. Ability to restrict data processing,

  4. Right to rectification,

  5. Right to erasure,

  6. Adherence to privacy laws

Data Ethics

Commit to ethical data collection standards, including, analysis and usage practices, ensuring that the data used is accurate, relevant, and collected with proper consent.

Inclusiveness

Actively address biases and disparities throughout the entire AI system lifecycle, by developing a framework to monitor, evaluate and design the AI systems through principles including;

  1. Non-discrimination and prevention of bias,

  2. Representative and high quality data

  3. Fairness

  4. Equality

  5. Inclusiveness in impact

  6. Inclusiveness in design

Accountability

Share accountability with Fundraising AI Actors for the AI applications that we develop, deploy, or utilize in the fundraising profession, ensuring that they align with our organization's or client's mission, values, and ethical principles that are;

  1. Verifiable and replicable

  2. Auditable

  3. Appealable

  4. Remediable

  5. Liable

Transparency and Explainability

Within reasonable efforts to safeguard corporate IP, will be transparent in the development, deployment, and utilization of AI technologies, providing, requiring or requesting clear explanations of AI methodologies, results, reporting, measurement, and potential impacts on participants. In addition, we will provide adequate visibility to consumers of our AI ecosystems outputs when autonomous AI has been utilized.

Continuous Learning

Commit to the responsible use of tested and untested resources while staying informed about the latest developments in Responsible AI, incorporating best practices into my work within the fundraising profession, and to share the responsibility of helping educate the broader fundraising community on Responsible AI best-practices.

Collaboration

Actively engage with and learn from my peers in Fundraising.AI, sharing my experiences, challenges, and successes in Responsible AI for fundraising.

Legal Compliance

Commit to being aware of, and abiding by, applicable laws, regulations, and best practices concerning AI development and operations pertaining to fundraising, data protection, and AI systems.

Social Impact

Strive to maximize the positive social impact of AI in fundraising while minimizing any potential harm by focusing on the needs of beneficiaries and communities.

Sustainability

Commit to considering the long-term sustainability and environmental impact of AI technologies and advocate for sustainable AI practices within my organization and the broader fundraising community.


Understanding the limitations of grant writing with AI

yellow and black road signby Breana Panaguiton

When using AI for generating ideas and content, it's important to remember that it has its limitations. For instance, AI cannot replace the expertise and knowledge of a human researcher, and it is not a good research tool.

Additionally, while the speed of content creation with AI can be useful, it can also create new opportunities to make mistakes, as the technology can generate large amounts of content quickly, but may not catch errors or inconsistencies.

Furthermore, keep in mind that AI has no grasp of real-world context, which can lead to inaccuracies or irrelevant content.

Finally, AI has no understanding of truthfulness or factuality, meaning that the content it generates may not always be accurate or trustworthy. As a result, it's important to use AI as a supplement to, rather than a replacement for, human expertise and judgment.

There are four limitations to keep in mind when using AI for grant writing:

  1. Generative AI is not a good research tool

  2. The speed of content creation creates new opportunities to make mistakes

  3. Generative AI has no grasp of real world context

  4. Generative AI has no understanding of truthfulness or factuality

Why isn’t AI a good research tool?

When you use generative AI to generate text, it's important to understand what's really happening behind the scenes. Essentially, the generative AI is making complex predictions based on probabilities.

Large language models analyze enormous datasets to learn patterns and relationships between words and phrases. They then use this knowledge to generate new text that is similar to the input data. The models are trained on huge datasets and use complex algorithms to predict what words or phrases are likely to come next in a given context. The model generates output by choosing words or phrases that are likely to follow the input, based on statistical probabilities.

It's important to keep in mind that the outputs created by generative AI are not based on actual understanding or comprehension of the content, but rather on patterns and relationships learned from the large dataset. While this can be useful for generating new ideas or content quickly, it's not a good substitute for in-depth research or analysis.

Generative AI still has limited reasoning abilities to evaluate sources, synthesize information, or draw conclusions based on a nuanced understanding of the subject matter. Additionally, generative AI may not be able to recognize biases or inaccuracies in the data it is trained on, leading to outputs that perpetuate these issues. As a result, it's important to use generative AI as a supplement to, rather than a replacement for, your own expertise and judgment when it comes to research.

🚨 It's important to keep in mind that the outputs created by the AI are not based on actual understanding or comprehension of the content, but rather on patterns and relationships learned from the large dataset.

Research tools

Generative AI

Best for

Finding and organizing information

Creating new content

Output

List of relevant results

Content generated based on probability

Accuracy

High with high-quality sources

Unknown without verification

Human role

Guidance, analysis and synthesis

Guidance, fact checking and editing

Example software

Google

Grantable

How does the speed of generative AI tools create new opportunities for mistakes?

Generative AI systems can produce pages of credible-sounding text in a matter of seconds. This is faster than most people can read, and a large quantity of AI-generated content requiring human approval can quickly pile up. Even those who initially review AI outputs vigilantly can be lulled into a false sense of security after a few minutes or hours of error-free products. It's crucial to remember that generative AI systems can inadvertently generate problematic content at any time.

A good way to think about writing with and without AI is to compare it to sewing by hand with a needle and thread versus using a sewing machine. When we write without AI assistance, we proceed only as quickly as the words come to us and we can type them. Similarly, with a needle and thread, we can only sew stitch by stitch, leaving us with plenty of time to guide the work.

Photo by J Williams on Unsplash

Writing with AI assistance increases the speed of output exponentially, just like a sewing machine. You are now guiding a machine that moves much more quickly than a person can. The advantage, of course, is that you can complete a project much more quickly using a machine to assist, while the hazard is that even momentary lapses in focus can immediately create larger problems or lead you further away from the goal of your work.

AI seems so smart, what do you mean it doesn’t understand the real world?

Despite their impressive abilities to comprehend and respond to a wide range of queries and prompts, generative AI systems have no grasp of the realities they're writing about. Don't be fooled by skillful references to relatively current events or insightful analyses of complex issues — these are all the result of complex mathematics behind the scenes.

Generative AI systems have processed unimaginably large data sets of text, from which they've discovered patterns of probabilities that connect words, phrases, concepts, and structures of the writing they've ingested. When you prompt one of these systems, it breaks your query down into numbers and probabilities and predicts each word that comes next.

At no time is the model checking with any validated sources of information to ensure it is pulling correct information for you. It's not pulling information at all; it's just predicting the next most likely word in a sequence.

At the same time, it is not trying to understand your query in the way another person tries to process the meaning of your words and what actions should follow. Generative AI models are simply turning your words into numbers and predicting which ones should come next.

What is an AI ‘hallucination’? Why does generative AI hallucinate?

AI hallucination refers to when a generative AI system produces outputs that are illogical, nonsensical, or even offensive due to the limitations of the software. Real-life examples may include entirely fictional people, places, events, and source materials.

When producing these hallucinations, the AI will sound every bit as confident as when the information is factual and correct. It is incredibly important that human users remain vigilant for such hallucinations when relying on AI-generated content in their work.

💡 Grantable is far less likely to hallucinate because the AI model is limited to using an organization's existing writing samples as source material to guide its writing.

A hallucination is not a malfunction; the AI system is performing exactly as it has been programmed to do. It has simply made a text prediction that we experience as being nonsensical or false. To the AI model, the text it has generated comes from a correct calculation of the probability that this sequence of words is an appropriate continuation of the human user's input query.


A view of the future of AI and philanthropic work

On November 30, 2022, OpenAI released ChatGPT, a research prototype aimed at studying user engagement with its large language models (LLMs). Within five days, ChatGPT attracted over one million users (15X faster than Instagram) and surpassed 100 million users in two months. The chatbot was able to generate remarkable responses to complex textual input, leaving the world captivated, amazed, and unnerved.


Computing advances, investments, competition between tech giants, and global awareness of LLMs are driving the fastest technological revolution in human history. Bill Gates, co-founder of Microsoft, compared this moment to the introduction of the graphical user interface, which made computing accessible to everyone. Sundar Pichai, the CEO of Google parent company Alphabet, has called this technology "more profound than fire or electricity."

LLMs can approximate the work of highly skilled knowledge workers and creatives in a matter of seconds and at a cost approaching zero. They are becoming more capable by the hour, and the quality of their outputs is improving just as quickly.

It is impossible to fully imagine how these systems will transform the world, and predictions range from armageddon to utopia. In a 2021 essay titled, "Moore's Law for Everything," OpenAI's CEO, Sam Altman, envisions a positive post-AI future, writing "As AI produces most of the world's basic goods and services, people will be freed up to spend more time with people they care about, care for people, appreciate art and nature, or work toward social good."

While we can't clearly envision many specifics about this new era in technology, we can reasonably assume that change will come more quickly than in the past. This is because the world is increasingly digitally connected, the speed of innovation is accelerating, as more people, resources, and prior breakthroughs compound, and AI dominance is becoming a major economic and geopolitical goal for man.

Artificial intelligence, like the systems we've been studying in this course, and even more powerful systems with artificial general intelligence (AGI), which can exceed human intelligence and capabilities, are being developed. Such technology creates the possibility of reaching unimaginably positive and/or terrifyingly dire outcomes depending on how humanity builds and regulates the use of these systems.

The nonprofit sector, or as I like to call it, the purpose-driven sector, must play a leading role in helping to embed human, planet, and justice-centered values in the technology itself and infusing policymaking with the wisdom of the sector that has long sought to remedy the brokenness of existing socioeconomic systems.

In day-to-day work, nonprofit sector professionals should stay engaged with AI tools and discussions both to further the mission at hand and to be fluent in conversations about creating responsible and ethical norms and governance for AI.


Self assessment

This short quiz is only meant to help you check your understanding of these materials. Your score is not recorded, so please write it down if you want to keep track for your own records.

Did this answer your question?