Waterloo, Iowa, Police Cautious About New Technology

Read Time:5 Minute, 2 Second

(TNS) – Artificial intelligence is making inroads to the field of law enforcement.

But machine learning has yet to make a big splash at local police departments.

“We don’t do any predictive policing or use artificial intelligence to predict where crime will occur. We don’t use any predictive analytics,” said Joe Leibold, chief of the Waterloo Police Department.

The agency does have a few AI tools at its disposal.

The Axon body-worn cameras issued to every officer include a transcription function, Leibold said.

Officers turn on their cameras when dealing with the public or handling calls, recording what happens with an unblinking eye. When downloading footage from the camera, officers can use the tool to automatically generate a written account of conversations and voices picked up by the microphone.

It’s designed to be a time-saving device, but Leibold said most officers bypass the technology and instead transcribe by hand for better accuracy.

Another time-saving device used by police allows investigators to quickly sort through hours’ worth of video to find the pertinent parts. For instance, if a car was stolen over the span of a weekend, an officer can program the software to flag sections of the video where action takes place instead of sitting through 48 hours of footage.

“You can skip to where there’s motion, but it’s all based on an officer saying I need to look at these videos. We don’t have anything running randomly,” Leibold said.

The Waterloo department also uses license plate readers, cameras that are attached to squad cars. The cameras are running in the background, passively scanning plates of passing vehicles and checking them against a law enforcement database. When it comes across the plates of a stolen vehicle, it will send an alert to the officer, who then can confirm the automated hit and take appropriate action.

“We’ve recovered several stolen cars with them. I can’t tell you how many,” Leibold said.

Plate data is also kept for short period of time before they are dumped, allowing authorities to check for the presence of certain vehicles that may become of interest later on.

For years, the city has been using stationary red light and speed cameras that photograph plates and issue citations for drivers committing those infractions at certain intersections around town. But those cameras don’t collect plate numbers of non-offending motorists, Leibold said.

Much of the downtown also has stationary surveillance cameras run by the city. Even the landmark Fourth Street pedestrian bridge has them. The cameras have been used to solve crimes, including the case of a man who was breaking into area businesses.

The cameras aren’t linked to a facial recognition system, but the quality of the image is there to allow a human view to recognize someone on camera, Leibold said.

Other technology is on the horizon.

One such system is predictive policing, which uses statistical crime data and incident reports to make recommendations on where authorities should focus their attention.

That technology sometimes faces a backlash. In 2020, the city of Santa Cruz, Calif., became one of the first municipalities to ban predictive policing — and facial recognition — after using it for about a decade, according to the Los Angeles Times . Critics said the technology was unfairly targeting minority communities, the Times article states.

Other departments have used AI to review officer body cam footage, looking for hints of bad police work or inappropriate behavior. The Waterloo Police Department does random body camera reviews without the aid of AI, selecting a certain percentage of videos for human review, Leibold said.

“We do human sampling. We have supervisors review random body cam footage to look for best practices,” he said. “Honestly, that’s a people business, in my opinion.”

Meanwhile, cutting-edge technology is being used locally to train police officers.

Hawkeye Community College’s police science program recently added virtual reality simulations to its curriculum.

“Nothing is pre-recorded. One hundred percent of the actions and vocals are controlled by an operator,” said Ben Scholl, law enforcement director at Hawkeye.

The Apex Officer system uses VR goggles and plastic handgun and taser mockups to walk officer cadets through a series of scenarios — traffic stops, hostage situations or just a person having a bad day.

There is a large assortment of settings — from back alleys to apartments to offices — populated with lifelike characters controlled by the instructor. The virtual characters’ mouths move when the instructor talks. They have a range of facial expressions and can pull out driver’s licenses or guns.

Scholl said the virtual reality system is more than a shoot/don’t shoot simulator. Communication skills play an important role in the drills.

“Some of them end up in force. Some of them don’t. It could go either way,” he said. “This is an area where we can identify training needs and also strengths.”

Scholl said the most requested training scenarios simulate mental health calls and First Amendment auditors — social media streamers who push the envelope filming interactions with police.

In the past, such lessons had been practiced with the college’s MILO simulator, where the characters and environment were projected on a flat screen in front of the officer as pre-recorded scenarios played out. Officers’ choices in the drills were limited, as were the characters’ options.

“There was no benefit to move on a screen. This starts to offset, so you can see the angles. They move around like they are supposed to, so it gives a more natural response,” Scholl said.

The Apex Officer simulator isn’t just for Hawkeye students. The college lends the system to local law enforcement agencies for ongoing training.

___