Artificial Intelligence: The Hotdog Conundrum
HBO’s Silicon Valley is a not-too-far removed parody of work at a small technology startup. In season four, Jian Yang creates a machine-learning app called SeeFood. The idea is that the app can interpret an image taken from a mobile phone and tell you, in text, what the food is.
Jian Yang works relentlessly on the app and is immensely proud to demonstrate the final product to his housemates. He begins by taking a picture of a hot dog, which the app correctly identifies. Everyone is elated—this will be the next big thing (“We’re going to be rich!”).
The twist comes just a few moments later when Jian Yang takes a picture of a pizza slice: “not hot dog.” His housemates are crestfallen. It turns out that Jian Yang’s artificial intelligence (AI) app is based on a classification prediction—so it recognizes everything as either a hot dog or not a hot dog.
The Technology Disconnect
Many of us have experienced that same disconnect between expectations and experience, particularly in the realm of AI. For as many delightful experiences we’ve had with the technology (autocomplete in Microsoft Office, for example), we’ve had 10x the disappointment in other ventures.
It’s easy to overpromise on the capabilities of AI in marketing materials (“Let’s throw some AI at the problem!”), leading to a disconnect between the promise of the technology and its actual utility to a customer.
Of course, technology alone doesn’t magically change your outcomes (in some cases, it can slow you down). Technology is magic when there’s close alignment between a real problem or pain point and the unique problem the technology is designed to solve. People, process, and technology must interact.
It’s essential to understand what AI is (marketing aside) and what it’s uniquely suited to do when exploring products to benefit your nonprofit or school. AI and machine learning predictions are—essentially—a collection of algebraic equations that articulate the most probable outcome to a very specific question. And the value of applying that AI will be determined by how much you care about the answer to that question.
Here are some best practices when evaluating products or capabilities in this space:
- Assess where your team is uncertain and identify in which aspect of the work that uncertainty is concentrated.
- Spend time brainstorming some specific questions that could be predicted to address this uncertainty (Is this a hot dog?).
- Articulate how removing the uncertainty would be valuable for the organization. What impact on the organization would you see as a result of less uncertainty in this area?
- What aspects of the team’s workflow could be made faster/shorter/more streamlined if that uncertainty was alleviated? Will this free up their time to focus on more valuable work?
“Hot dog” or “not hot dog” may seem silly. But what if it wasn’t a hot dog that Jian Yang’s app identified—but breast cancer indicators on an MRI reading? What if artificial intelligence drove the cost down of early identification to a point where it is accessible to more people, meaning we can catch more cases in earlier stages?
For another resource on thinking through the potential value of AI in a specific situation, check out the AI Canvas Framework created by Ajay Agrawal, Joshua Gans, and Avi Goldfarb on Harvard Business Review.
Blackbaud’s Artificial Intelligence Journey
AI is great when its purpose aligns with the outcomes our customers hope to achieve. But when it is not, it’s just noise. That’s why we are so excited to be providing industry-specific AI-powered tools that will create real impact for our customers.
In June 2022, we released Prospect Insights—a new software tool within Blackbaud Raiser’s Edge NXT® that enables social good professionals to access actionable, AI-powered insights to drive major giving.
At Blackbaud, we are on a journey to build intelligent experiences (inclusive of AI-enabled ones) that are truly valuable to our customers. We have a team that deeply understands the social good sector and the problems nonprofits and schools face every day, and we come to the table with that in mind.
Over the coming weeks, I’ll share some of what I’ve learned on my journey into AI over the past few years, and I hope you’ll join me. Cheers!