Search blog

Harnessing generative AI for good

A man on a laptop with message bubbles being held by a robotic hand.

Generative AI has captured our collective imagination and will likely transform the world as we know it. While the tech sector pours in tens of billions in a race for dominance and future profits, how should nonprofits and philanthropists be thinking about our role?   

Before we get completely swept up, let’s not forget the previous technology hype cycles that have come and gone. The unfulfilled promises of blockchain to advance social good, a proliferation of mobile apps in developing countries that failed to garner adoption, and cheap laptops that were going to change the world are just a few examples. To be clear, I believe there is far more substance and potential with generative AI. But these should nevertheless serve as a cautionary tale. 

To help nonprofits and philanthropists thoughtfully assess the risks and opportunities generative AI poses for our collective efforts to ensure all people and our planet thrive, I suggest three grounding principles:   

  1. Fall in love with the problem, not the technology 
  1. Fast follow 
  1. Keep equity front and center 

1) Fall in love with the problem, not the technology

There’s an adage in innovation: Fall in love with the problem, not your solution. This equally applies to technology. Rather than getting swept away by the allure of generative AI and casting about for some possible application (like a hammer looking for a nail), we should start instead by deeply understanding the problem we aim to solve from the perspectives of those directly impacted. 

Working with emerging technology like generative AI entails a high degree of uncertainty, and thus requires a different approach from the implementation of well-defined programs. By staying human-centered and running small, iterative experiments, we can more quickly identify where the greatest potential for real value for real people in the real world lies. What we need most is agility—the ability to learn and pivot as quickly as possible. 

2) Fast follow

The Gartner Hype Cycle captures the euphoria new technologies tend to generate in the early days. New technologies often disappoint initially, but over time, a more grounded reality emerges where we begin to reap tangible benefits.  

While other types of artificial intelligence have long been productively integrated in our daily lives (including extracting, categorizing, and making sense of data here at Candid), the recent surge of interest in generative AI following the launch of Chat GPT 3.0 has likely put it near the peak of the hype cycle. Given both the financial resources and technical talent in the tech sector overwhelmingly dwarfs that of the social sector, tech companies are clearly best positioned to work out the very real kinks (such as disturbing hallucinations). Rather than stumbling through all the same expensive lessons, you’ll likely reap far greater benefit for far less cost as a fast follower. 

When I was at Google, every few years a new search site would arrive on the scene that appeared to deliver similar or better results. But once the site started to get real traction, it would catch the attention of spammers and its quality would rapidly decline. Flashy demos can quickly fizzle when faced with the complexities of the real world. One way I’d gauge when generative AI is ready for primetime is when it begins to replace core business functions like the home page of Google.  

3) Keep equity front and center

Technology generally has a tendency to exacerbate inequities, as those who have been historically marginalized often don’t have the resources or capacity to fully benefit. Consider the persistent gender gap in access to both mobile phones and the internet. We have the responsibility to ask: who is generative AI going to benefit most? 

Today, generative AI is being subsidized by venture capitalists and big tech, so access has been free or cheap. But training and running large language models is expensive. If we believe Bill Gates that someday we will have AI personal agents to write for us, plan for us, and do work for us, who will be able to access and afford them? Will low-income countries be left further behind? 

There’s a crucial role the social sector can play to get in front of the potential downsides of technology. For example, Candid’s Foundation Directory helps nonprofits identify potential sources of funding. To ensure this powerful tool benefits all nonprofits, not only those who are already well-resourced, this year we made Foundation Directory free for smaller nonprofits; stopped charging for virtually all our trainings; and established a new department focused on creating a level playing field for all nonprofits to access funding. 

Ensuring everyone, regardless of background, education, or economic status, can fully benefit from generative AI will be central to whether this revolutionary technology advancement exacerbates or reduces existing inequities. Yes, new applications are sexy and garner press coverage. But paying attention to who benefits most may, in the long run, result in the biggest impact. 

Looking forward

Applications based on generative AI hold the potential to drive broad and far-reaching impact across our society. Could it revolutionize personalized education, expand access to quality healthcare, or empower environmental forecasting for climate change mitigation? The full realization of this promise will take time, but in the long run our society will likely be fundamentally changed. If we can harness generative AI for good, the potential is limitless. 


Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  • Caroline says:

    December 5, 2023 1:16 pm

    Very explanatory. Thanks!

  • wdqwa says:

    August 17, 2023 5:03 pm