Elizabeth Smart speaks in Barron, Wis., March 15, 2019.

Photo: Renee Jones Schneider/Zuma Press

Are you ready for a good story about the internet, one that doesn’t mention Twitter’s reality-censoring cesspool of snark or twerking TikTokers? About the internet’s power to connect all of us instantly? I sure am.

I recently had a Zoom call with Elizabeth Smart—yes, that Elizabeth Smart, who in 2002 was abducted as a 14-year-old and held for nine months in Utah, a harrowing experience for her and her family. I once lost a son at Disneyland for three minutes and went out of my mind.

Ms. Smart not only is an advocate for missing children but is doing something about it. She told me 600,000 children go missing in the U.S. every year and that 98% of missing children are located within a few days. She noted that “after 48 hours, the chances of being found are almost zero, but obviously not zero in my case.”

Ms. Smart had good things to say about the Amber alerts posted on highway signs and blasted to cellphones, but pointed to their shortcomings. There is an age limit, and law enforcement needs a reasonable belief of imminent danger, so it is used rarely. Something else is needed. She connected with an Oregon-based tech company to help design and promote its Q5id Guardian app, which makes use of crowdsourcing. When a child is missing, the app sends out localized alerts to a network of volunteers, all verified to avoid potential predators on the system. “Remember,” she said, “I was found by everyday people paying attention, not law enforcement.”

The Guardian app launched in November. I asked Ms. Smart what makes her think it can be scaled to an effective size. “I believe in the goodness of people.” Amazing given her experience.

That goodness does exist. Wikipedia was enabled by digital crowdsourcing in 2001, but real-time volunteer networks started gaining traction in March 2014, when Malaysia Airlines Flight 370 from Kuala Lumpur to Beijing never showed up. The last radar contact was over the South China Sea. CNN had blanket coverage for what seemed like months. Soon eight million volunteers from around the globe began poring over satellite images covering 130,000 square miles of ocean on their computers. They never found it, but the effort was astounding.

A few years back I met Israeli entrepreneur Eli Beer. As a youth, he volunteered as an emergency medical worker and watched a child choke to death before an ambulance stuck in traffic showed up, even though a doctor lived in the same building as the child. So he created United Hatzalah, a crowd of volunteers who respond to emergencies, often on bikes or “ambucycles” loaded with medical equipment to avoid traffic. There are more than 6,200 volunteers in Israel who can respond within 90 seconds to medical emergencies and terrorist attacks in their neighborhoods. This augments, not replaces, ambulances. Since 1989 they have helped 5.5 million people. It is expanding around the world, including in Ukraine.

More crowdsourcing: After the Golden State Killer was identified in 2018 using DNA evidence to narrow down his family tree, amateur DNA sleuths began doing the tedious work of digging through genealogies. The DNA Doe Project enables volunteers to scour GEDmatch, the free genealogy service this column has written about. These volunteers try to identify “John and Jane Does,” often cold cases of missing people or crime victims. In April 2018, they helped identify a murder victim known as the “Buckskin Girl” and have since identified more than 60 others.

I was once pitched an app that would warn users about sketchy neighborhoods in various cities. It combined police reports and crowdsourced inputs to warn the app’s users, mostly single women, where to avoid walking. It never got off the ground because many early reviews called it racist. Sad state of affairs.

You may have heard the recent hullabaloo over the machine-learning-enabled service ChatGPT from OpenAI. You ask it a question and it can offer an answer in rather eloquent prose. This is because it scans billions of pages of prose written by humans; think of it as nonvoluntary crowdsourcing. Very cool. Strangely, the answers are often erroneous. Eloquent, but wrong—maybe in need of another crowdsourced human effort to correct them. I worry these AI apps will soon be trained with the output of other AI apps—garbage in, garbage out—proving the value of real humans again.

For Ms. Smart and the Guardian app, I suggested that what she really needs is access to a broader technology platform to get emergency information out when needed. Google, Apple and others know where we are. I doubt they would agree to 600,000 alerts every year, but we need something more than rare, centralized Amber alerts. Let’s hope this works. Great things happen when you believe in the goodness of people.

Write to kessler@wsj.com.