Good News Saturday: Five Ways AI Is Helping

My last few posts were about governance, agent risk and the end of software gatekeeping. Important topics, but not exactly weekend reading. So here is a new series: Good News Saturday, a short look at places where AI is quietly making things better.

1. Catching pancreatic cancer years earlier

Pancreatic cancer is deadly mostly because it is found late. In April, Mayo Clinic reported that its AI model (REDMOD) can spot early tissue changes on routine abdominal CT scans up to three years before a clinical diagnosis. The scans were already being taken for other reasons. The AI just looked more closely.

Mayo is now testing it prospectively in higher-risk patients, which is exactly the right next step.

2. Finding breast cancer sooner

A Swedish study in Radiology found that commercial AI systems flagged early signs of breast cancer on mammograms up to six years before diagnosis in some patients. Separately, an NHS study of about 175,000 women found that using AI as a second reader raised detection from 7.54 to 9.33 cancers per 1,000 women, with fewer false positives and less reading time.

3. Giving a voice back to people with ALS

UC Davis neuroscientist Sergey Stavisky won the 2026 Chen Institute and Science Prize for AI Accelerated Research for a brain implant that turns attempted speech into a synthetic version of the patient’s own voice. For one man with ALS, it reached over 99% word accuracy and carried more than 2.7 million words over two years.

That is not a demo. That is someone talking to his family again.

4. Listening for orcas

Only about 76 southern resident orcas are left in the Salish Sea, and ship noise makes it harder for them to hunt. OrcaHello, an open-source tool that started at a 2019 hackathon, listens to underwater microphones 24/7 and flags orca calls. A human expert confirms each alert before it goes out. The Port of Seattle uses the alerts to pause pile-driving, and Washington State Ferries slow down when the whales pass.

No new hardware, no tagging the animals. Just better listening.

5. Flood warnings where there were none

Many of the most flood-prone places never had early warnings because they lacked local river data. Google’s Flood Hub uses AI to forecast river floods up to seven days ahead in 80 countries covering about 460 million people. In Bihar, India, an evaluation found that communities receiving the alerts had 30% lower medical costs after floods.

In March, the service added urban flash flood forecasts. That matters: flash floods cause roughly 85% of flood deaths worldwide.

The common thread

In every one of these, a person is still in the loop. A radiologist reads the scan. A clinical team runs the implant. An expert confirms the orca call. A national agency issues the flood warning. The AI finds the signal faster than people could alone, and people decide what to do with it.

That is the same model I keep coming back to in regulated work: AI as a tool, humans at the helm. It turns out it works for whales, too.

Have a good-news AI story? Send it my way on LinkedIn and it may show up in a future Saturday post.