CHICAGO (AP) — Michael Williams’ wife begged him to remember their fishing trips with the grandkids, the way he used to braid his hair, anything that would bring him back to his people outside the concrete walls of the Cook County Jail.
His three daily calls had become a lifeline, but when they dwindled to just a few a week, the 65-year-old felt he couldn’t go on. He planned to kill himself with a stash of pills.
Williams was arrested last August, charged with the murder of a young man from the neighborhood who asked him to drive him during a night of unrest because of police brutality. The main evidence came from a video of a car driving through an intersection and a loud noise picked up by acoustic sensors. Prosecutors said audio technology powered by a secret algorithm indicated that Williams shot and killed the man inside his car.
“I kept trying to figure out, how can they get away with using technology like that against me?” said Williams. “It is not fair.”
Williams was jailed for nearly a year before prosecutors, citing insufficient evidence, asked a judge to drop the case.
Williams’ experience highlights the real impacts of society’s growing reliance on algorithms to help make consequential decisions about public life. This is particularly evident in law enforcement, which has turned to technologies such as acoustic gunshot detection. One such company, ShotSpotter, says its evidence is gaining acceptance in courtrooms, now some 200. ShotSpotter’s website says it is a leader in law enforcement technology solutions that help stop gun violence by using algorithms to classify 14 million sounds as gunshots or something else.
But an Associated Press investigation, based on thousands of internal documents, emails and confidential contracts, as well as dozens of interviews, identified serious flaws in ShotSpotter’s use of evidence in court.
AP’s investigation found that the system can miss actual gunfire right under its microphones, or misclassify sounds of fireworks or overturning cars as gunfire. ShotSpotter forensic reports have been used in court to falsely claim that a defendant fired at police or provide questionable counts of the number of shots fired.
ShotSpotter touts its algorithm-based technology as virtually foolproof. But its algorithms are a trade secret, largely impenetrable to the public, jurors and police oversight boards.
The company identifies possible shots with the acoustic sensors. Then, ShotSpotter employees listen to the audio recordings of those sounds and either confirm or change the source of the sounds, introducing the possibility of human bias. Employees can and do change the location or number of shots at the request of police, according to court records. And in the past, city dispatchers or the police themselves could make some of these changes.
Amid a national debate over racial bias in policing, civil rights advocates say the criminal justice system shouldn’t outsource some of society’s most important decisions to computer code.
ShotSpotter CEO Ralph Clark said the AI details are “not really relevant”.
“The point is, anything that’s ultimately produced as a gunshot has to have eyes and ears on it,” Clark said. “Human eyes and ears, okay? »
This story, backed by the Pulitzer Center for Crisis Reporting, is part of an ongoing Associated Press series, “Tracked,” that investigates the power and consequences of decisions made by algorithms on people’s daily lives.
Police chiefs call ShotSpotter a game changer. The technology has been installed in about 110 US cities, often disproportionately placed in black and Latino communities. Law enforcement officials say it helps get officers to crime scenes faster, making their neighborhoods safer.
But university researchers who examined 68 major metropolitan counties from 1999 to 2016 found that the technology did not reduce gun violence or increase community safety.
On a Sunday evening in May 2020, Williams said Safarian Herring, a 25-year-old man he said he saw in the neighborhood, waved at him for a ride. Williams told police a vehicle pulled up next to him and someone shot Herring.
“I was yelling at my passenger ‘Are you okay?'” Williams said. “He did not answer.”
He rushed to the emergency room. Herring died a few days later.
Three months later, the police showed up and, after questioning, charged Williams with first-degree murder.
“When he said that to me, it was like something inside me had just died,” Williams said.
On the night of the shooting, ShotSpotters sensors identified a loud noise that the system initially attributed to 5700 S. Lake Shore Dr., according to an alert the company sent to police. This material anchored prosecutors’ theory that Williams shot Herring inside his car, even though the supplemental police report did not cite a motive, mention eyewitnesses or a recovered weapon.
Prosecutors also relied on surveillance video showing Williams’ car running a red light, as did another car that appeared to have the windows open, ruling out that the shot came from the passenger window of the another car, they said.
Chicago police did not respond to AP’s request for comment. The Cook County State’s Attorney’s Office said in a statement that after careful consideration, prosecutors “have concluded that the totality of the evidence is insufficient to discharge our burden of proof.”
As ShotSpotter’s gunshot detection systems expand across the country, its use as evidence in courtrooms has also grown, including 91 cases over the past 4 years.
“Our data compiled with our expert analysis helps prosecutors hand down convictions,” said a recent ShotSpotter press release.
But as the cases increase, defense attorneys are becoming increasingly skeptical.
During testimony in 2016 in a shooting trial involving a Rochester, New York officer, ShotSpotter engineer Paul Greene said an employee reclassified sounds from a helicopter to a bullet because the Rochester police had told him.
In the Williams case, evidence from the preliminary hearings shows that ShotSpotter initially said the noise picked up by the sensor was a firecracker, but a ShotSpotter employee renamed it a gunshot.
Later, ShotSpotter senior technical support engineer Walter Collier changed the reported address of the sound in Chicago to the street where Williams was driving, about a mile away, according to court documents. ShotSpotter said Collier corrected the report to match the actual location identified by the sensors. The company then provided AP with a copy of the full real-time alert, which contained a mailing address, location maps, and latitude and longitude coordinates, as did the second detailed forensic analysis prepared by Collier. The assigned civic address changed from the first to the second report, but the location identified on maps and GPS coordinates in both reports remained around the same intersection.
Last month, a judge dismissed the case against Williams.
ShotSpotter insists it has warned prosecutors not to rely on its technology to detect gunshots inside vehicles or buildings, citing language in its $33 million contract with the service. Chicago police.
Williams’ attorney, Brendan Max, said prosecutors never shared this critical information.
Williams remains shaken. As he walks around the neighborhood, he searches for the acoustic sensors that nearly landed him in prison for life.
“The only places these devices are installed are in poor black communities, nowhere else,” he said. “How many of us will find ourselves in this same situation?”
This story was originally published on August 19, 2021. The Associated Press reported that a ShotSpotter engineer changed the reported Chicago address of a sound the company said was a gunshot on the street where Michael Williams was driving. The story included ShotSpotter’s explanation that the engineer corrected the mailing address generated in his initial real-time alert to match the actual mailing address identified by the company’s sensors. The company has now provided the AP with a copy of the full real-time alert. The two reports released by the company – the initial real-time alert and the detailed forensic analysis later filed in court – contained a mailing address, location maps, and latitude and longitude coordinates. The assigned civic address changed from the first to the second report, but the location identified on maps and GPS coordinates in both reports remained around the same intersection. ShotSpotter says the street address in the initial real-time alert sent to police was wrong because the GPS coordinates were in a large park for which the officially designated address was about a mile from the actual location identified by the sensors.
Mendoza reported from Newark, Calif. Associated Press writer Roselyn Romero in San Luis Obispo, Calif., contributed to this report.
Contact the AP Global Investigation Team at [email protected]