Tuesday, February 25th

IN DATA NEWS

Judge Throws Out Facial Recognition Evidence In Murder Case

An Ohio judge throws out facial recognition evidence in a murder case, preventing prosecutors from securing a conviction. Clearview AI, a controversial facial recognition software, was used to identify the suspect from surveillance footage.

“With no immediate leads, investigators turned to surveillance footage taken six days after the crime. They used Clearview AI, a controversial facial recognition software, to identify a suspect…Acting on this identification, police obtained a search warrant.., where they recovered a firearm and other evidence. But as the trial approached, a flaw in the investigation came to light: police had not independently corroborated [the subject’s] identity before executing the search warrant, nor had they disclosed the use of facial recognition in their affidavit.”

Did you pick up on the questionable sequence of events?

1. The surveillance footage was reviewed SIX days after the crime. Now maybe the investigators needed to obtain approval in order to access this footage, but SIX days seems like a mighty long time.
2. Using Clearview AI was simply a boneheaded move. The law enforcement department should’ve rescinded its use. But they didn’t. A quick interwebs search would’ve revealed that Clearview AI had been at the center of many lawsuits, including ones in Australia (2021) and the United States (2022).
3. The investigators didn’t independently corroborate [the subject’s] identity. 🤦🏾‍♀️Well, a 1-hour mandatory workforce development seminar and internal memo could’ve prevented this stumble.
4. The investigators didn’t tell the appropriate authorities that they used Clearview AI. Again, 🤦🏾‍♀️ Yet another 1-hour mandatory seminar that could’ve avoided that whole embarrassment.

So what’s the takeaway? VERIFY AI’S OUTPUT. EVERY. TIME. Facial recognition systems misclassify light-skinned men 0.8% of the time as compared to 34.7% of the time for darker-skinned women. We know this fact due to the 2018 Gender Shades research study by Drs. Joy Buolamwini and Timnit Gebru.

It’s astonishing that in 2024, when this case was being investigated, that no one within the local law enforcement and criminal justice system raised questions about how the identification was made prior to the search warrant. Not the supervisors of the investigators. Not the lawyers who wrote the search warrant. Not the judge that signed said search warrant.

But wait, yes we know why this happened. An unsolved open case doesn’t look as good as a solved closed case. Also, no data provenance questions were asked as the identification of a Black man as the offender remains socially accepted and expected.

Glad this judge threw out the evidence. Maybe the investigators, their supervisors, lawyers and other judges learned their lesson: VERIFY.
​

​
​Like what you're reading? Find it informative and insightful? You can sponsor the Rebel Tech Newsletter and follow us on LinkedIn.

​

​

HAPPENINGS & APPEARANCES

• [🎥 MITIGATING BIAS WEBINAR REPLAY 🎥] If you missed the McGraw-Hill AccessEnginning webinar for Mitigating Bias in Machine Learning on February 12th, don’t fret! Check out the replay here. Drs. Berry and Marshall were joined by two of the chapter author contributors, Jazmia Henry and Dr. Thema Monroe-While. The discussion gave an overview of the textbook and described ways that disparities show up digitally. Hear the insightful questions from attendees and thoughtful responses provided by the panelists.
• [✅ RESPONSIBLE DATA STRATEGIES ✅] Check out the 10 simple rules for building and maintaining a responsible data science workflow. No matter the stage in your data pipeline, there’s a rule you can practically implement to help mitigate risks. Learn more HERE.
• [✨DATA COURSE ✨] Available on LinkedIn Learning, you can get a snackable overview or refresher on data modeling by taking Practical Database Design: Implementing Responsible Data Solutions with SQL Querying. Get started HERE.

LAUGHING IS GOOD FOR THE SOUL

​

Stay Rebel Techie,

Dr. Brandeis

​

Thanks for subscribing! If you like what you read or use it as a resource, please share the newsletter signup with three friends!