-
Experimenting with AI: Potential Benefits & Pitfalls
Many nonprofits continue to explore and consider the potential uses of Artificial Intelligence (AI) in their work as organizations continue to determine ways to achieve efficiency and maximize staff time. But with this transformative technology comes many big risks and much responsibility around ethical use, as cited in this article: https://www.philanthropy.com/article/how-nonprofits-can-use-a-i-well-and-avoid-pitfalls
As experiments with AI in nonprofit operations continue, understanding the positive and negative results (both intended and unintended) will be critical for the sector.
To that end, we want to hear from YOU! How are considerations around the use of AI coming up in the context of your fiscal and other operational functions? How have you used it to date? How would you like to use it? What are your biggest concerns around using it? And if you have used it, what have been the biggest benefits and pitfalls?
-
2 Replies
-
I am a longtime AI leader, starting with founding for-profit machine learning companies in Silicon Valley, and then bringing technology (including AI) to multiple areas of the nonprofit sector. I just want to point out that at least 90% of the things I hear AI is going to do in the nonprofit world are going to fail over the next three years. It won’t be as bad as blockchain (0 for 1000 over the last seven years of the hype in terms of social impact at scale), but the vibe is similar.
Many of us in the tech sector make fun of generative AI, with terms like “stochastic parrots,” spicy autocomplete,” and “spell-checkers on steroids” (this last one being mine). The technology is fascinating, amazing and still dumb as a brick. It does not understand what it is saying. The amazing thing is that it is still useful. Just not miraculous. And when expectations are set unreasonably (lay off half your development team!), leaders are going to be disappointed.
One courageous leader recently shared his story of disappointment at an international meeting. His national crisis helpline had spent 50,000 (roughly dollars) on a simple chatbot (no AI). It did miracles: his team managed to deal with a several-fold increase in case loads thanks to the time savings. Over the last year, his team spent 1,000,000 (roughly dollars) on an AI-powered chatbot. And its performance was sufficiently weak they did not deploy it.
As a technologist (and former CFO for 10+ years), when someone comes to me with a jazzy technology and is trying to find out what to use it for, I suggest they push the pause button. What are their top three challenges where tech might plausibly make the humans in the system more effective? Then, what are the most affordable, supportable, trainable, sustainable tech solutions which meet the need. It might be just a simple boring piece of well-understood tech, rather than the latest blockchain/generative AI hype!
-
I’ve gotten good mileage out of AI (while also taking my two grains of salt). It has helped me to write, brainstorm, and proofread. It generated great images for me to use on my blog.
It’s wonderful for writing the first draft of an email, especially those difficult to write ones like “how can I politely decline this meeting that is useless to me.”
