This actually sounds pretty reasonable to me.
With thousands of hours of body camera footage and non-emergency calls and not enough staff to tackle it all, Harris County Sheriff’s officials are thinking of turning to artificial intelligence to handle some of the load.
Officials with the office named finding ways they might use AI as one of their chief goals for 2024, becoming one of the agencies on the forefront of the technology in law enforcement. But while some agencies are turning to technology like facial recognition software, administrators at the sheriff’s office said they see the biggest benefits are AI’s ability to parse through vast quantities of data.
“We are exploring the potential of AI to enhance public service by facilitating faster and more efficient data retrieval on our website,” said Gary Spurger, director of information technology for the sheriff’s office.
The sheriff’s office has not any contracts for the technology, but are testing its use in several different areas, according to Spurger. Those include using it to sort through body camera video and helping people find incarcerated family members.
Spurger gave an example of a deputy looking into a case involving more than 100 hours of body camera video to sort through. The ability to enter keywords, such as “man wearing blue jeans and a T-shirt,” will help the investigator more efficiently sort through data.
But while the technology can be a good aid, it would ultimately fall on employees to make final determinations, Spurger said.
That to me is a perfectly fine use case for AI. It’s in line with how we use AI in a cybersecurity context, as a way to do refined searches of vast quantities of data with a goal of identifying a generally small number of items of much greater interest. It’s still on the human analyst to review the search results and determine what warrants further investigation. Perhaps this what HPD meant when it was talking about using “AI surveillance” in its operations, and if so then I mostly retract my concerns. AI is good for taking labor-intensive drudge work and turning it into manageable tasks. If that’s the direction they’re going, and assuming that sufficient oversight and checkpoints are in place to assure that algorithms aren’t making judgment decisions, it’s worth pursuing.