AI at Candid: Powering technology to support the sector’s lasting success
It feels like 2023 has become the year of generative artificial intelligence (AI). When ChatGPT launched late last year, it widely became an instant sensation for its uncanny abilities to produce human-like text.
Generative AI products, such as ChatGPT, are examples of a branch of AI called machine learning, which is concerned with learning from data to surface trends and predictions. An active field of research in the 1960s, machine learning is now deeply embedded in our daily lives—from receiving recommended products while we’re online shopping to Google helping us autocomplete our emails.
As these technologies mature, so do their applications. Modern uses of AI, such as machine learning, include understanding text, recognizing people and objects in images, and figuring out who you might know on social media platforms.
In this blog, I will first explain how Candid currently harnesses the power of AI technologies in our data and tools. Then, I will share our vision for how we will use generative AI to build future solutions that lend further value to the social sector.
A glimpse into AI at Candid
When considering the applications of AI and machine learning at Candid, we keep the following questions in mind: Does this application help our staff do their work more effectively? Can this tool help people arrive at a deeper understanding of how Candid’s data can support their success?
Here are just a few ways we leverage AI and machine learning in many of Candid’s current products and data enhancement processes.
Turning data we collect into the most comprehensive view of the social sector. Candid uses machine learning to automatically tag new grants, request for proposals (RFPs), and mission statements to our Philanthropy Classification System (PCS) taxonomy—or our way of organizing and enriching data to make it usable. Using AI in this way allows us to add millions of data points each year that provide a highly detailed view of activity in the sector. It also helps ensure people can access the robust and holistic information they need to advance their missions.
Making searching for funders and grants fast and easy. If you’ve ever run a search in Candid’s Foundation Directory, you’ve already interacted with our AI-driven PCS auto-classification capabilities. Powering Candid’s search technology with machine learning helps people quickly find information on potential funders and grants. This means nonprofits can spend less time sifting through the vast array of funding opportunities available and more time pursuing those that are most likely to be aligned with their missions.
Helping the sector stay up to date with the latest information. Candid’s social sector news feed uses machine learning to analyze news articles and press releases for new grant awards, requests for proposal (RFP) announcements, and identify the relevant organizations, people, and geographic locations associated with each. With AI, Candid makes the latest social sector information actionable in real time to individuals and organizations seeking funding, donors performing due diligence, and researchers tracking current social sector activity.
Facilitating intelligent funding connections (coming soon.) To help fulfill Candid’s mission to get you the information you need to do good, we are testing a new tool to connect nonprofits with potentially relevant funders. This tool will not only use AI, but it will also tap into the rich networks of social sector relationships emergent from Candid’s historical grantmaking data set, including nearly 30 million grants for over 2.5 million nonprofits worldwide.
Where is AI at Candid going?
To build the next wave of Candid’s AI technology, we will make use of generative text capabilities. Our initial plans are aimed at freeing up the time and resources nonprofits spend finding funders, attracting donors, and securing grants, to better focus on advancing their missions. Here’s a sneak peek into what awaits.
Generating first drafts of grant proposals. Our current work involves experiments on generating first drafts of grant proposals for nonprofits that are actively prospecting for funding within our tools. This may be especially useful for nonprofits responding to RFPs as they are announced. The more information we know about your nonprofit, the more detailed and specific first draft of a grant proposal our future AI tool will be able to generate. By updating your Candid profile and sharing your organization’s programs, budget, and demographic information, we can better support your future grant-seeking efforts.
Making it easier to attract donors online. We constantly look for ways to make it easier for nonprofits to claim and update their Candid profiles. A cross functional team at Candid is investigating how generative AI can use an organization’s recent tax forms to automatically prepopulate nonprofit profiles. Additionally, to make nonprofits even more attractive to potential funders, can generative AI be used to fix common text formatting issues (e.g., punctuation and capitalization) from IRS Form 990 fields before being displayed on Candid profiles? This is just one of the many questions we aim to answer to better serve the needs of the social sector.
As exciting as the possibilities of generative AI are, we are proceeding carefully to avoid some of the troubling behavior which can arise from this technology. As advocates of responsible and effective adoption of new technology for the greater good, we are excited to develop advanced tools for people to access even more actionable information from a vast social sector.
We strive to improve the AI-supported tools we build at Candid, and we rely on feedback from our users. If you have ideas for how we can continue to improve Candid’s products, please share them in the comments below.