Pam Lafave goes to the Wednesday evening session at Doc Arra’s museum knowing there isn’t much chance of anyone wanting to work after the latest developments in Martha Crawford’s case. After hearing the gossip in town during the weekend she decides to read Mary Jane Bloom’s column for once. She’s surprised to see it supports the claims of the gossipers. As she expects, she finds Mary, Joe, Peter, and Sam already sitting around the dining room table discussing the case. Peter says, “I’m glad the rapist left our area. Now, Sarah can go out again without fear of being raped.”
“My parishioners were relieved this Sunday. I hope it’s true,” says Joe.
“You believe the stories, Peter? Most people at the Courthouse are skeptical,” says Mary Anderson.
“I’m happy Sarah is no longer fearful, Peter, but I don’t buy the gossip or Mary Jane Bloom’s column. I suspect the sheriff is at a dead-end in the investigation. He may be spreading positive rumors to dampen down fears,” says Pam.
Sam finally comments, “When I heard Deputy Wilson telling the story Thursday night I didn’t believe it. I concluded the sheriff is at a dead end again like you said, Pam. Remember, I said I had an idea I needed to research? I spent the weekend researching the idea. It isn’t my idea, I heard about it from my nephew, Jim. He’s a young professor in the Criminal Justice Department at the Dearborn campus of the University of Michigan. I went to see him Saturday and Sunday. If we want to continue to search for Martha’s killer there’s an approach we may be able to use.”
Mary says, “I want to find that rapist. I read the sheriff’s file and the medical examiner’s report in case we decided to investigate. I wanted to be ready to help if there’s anything I can do. Tell us about the approach, Sam.”
Seeing Pam nod in agreement, Sam says, “Okay, sit back and relax. It’s new and takes explaining. The way my nephew explained it to me is the best way to understand it. We have all experienced seeing ads pop up on social media or in our email if we don’t filter out spam. If you are like me sometimes these ads appear when I have been looking for something online, or sometimes when I’ve run out of something. It’s like someone is reading my mind and sending me specifically targeted ads.”
Joe interrupts saying, “I hate that even though I am usually interested in the product or I need it. How do they know?”
Sam continues, “Jim says companies like Facebook and Google that depend on ad revenue have advanced software monitoring the web sites we visit, the products we buy, our location and even our emails. They use sophisticated artificial intelligence algorithms, based on sociology and psychology, to analyze the data they collect and forecast what product ads we’ll likely respond to.”
“That seems logical enough, even if it’s irritating,” says Peter thoughtfully, “how does it help us search for Martha’s rapist?”
“Here’s where it gets interesting,” says Sam, “There’s a man in Jim’s department who pioneered the development of much of the artificial intelligence-based software used for the targeted ads. He’s an expert in artificial intelligence research and has degrees in psychology. He wanted to use his expertise for a higher purpose than advertising. He left Silicon Valley for the U of M with an idea for using the technology he created to catch dangerous criminals.”
Pam says, “Now you’ve got my attention. How does that work?”
“Let me continue,” replies Sam. “His idea is to combine monitoring the data used to target ads with state and national criminal databases, with public databases used by companies to evaluate potential employees and photographic data from the surveillance cameras being used in our cities and businesses. Then the social and behavioral sciences, the physical sciences, statistics, and the humanities are all considered to forecast criminal behavior. The goal is to predict the probabilities of when and where a known criminal will commit the next crime. He also hopes it helps locate known criminals.”
Mary asks, “Can you give us an example of how it works?”
Sam replies, “I think so. Suppose the Detroit police are after someone committing a series of robberies. There are two approaches. If they suspect a known robber but don’t know where or when he’s likely to strike again, the AI-based software analyzes all the available data on the suspected robber and outputs two graphics. One is a set of concentric circles overlaying a city map. Each circle represents the probability the suspect will commit a robbery within the enclosed area. The second graphic is a simple timeline of the probability the suspect will commit a robbery. The police then may see there’s an 80% probability the suspect will commit a robbery at a convenience store in a certain neighborhood and a 70% probability it will happen within a three-day window the following week. They can either focus patrols in that neighborhood during those three days or alert the convenience store owners or do both.”
“I see how that could help the police. What’s the other approach?” asks Joe.
“They look at the graphics for all known robbers. This will show the likely hot spots for robberies and when the robberies are most likely to occur. This data will allow the police to more effectively deploy patrols,” says Sam.
Pam says, “I wish we had that software in Grand Rapids when I worked as a detective. We could have systematically caught most of our habitual criminals.”
“That’s what the professor wants to make possible,” says Sam.
Mary says, “Now I see how he’s trying to catch known criminals. How does that help us search for Martha’s killer? We already know when and where she was killed. How does that relate to her killer?”
“I think I know,” says Pam. “If we can run the software for all known rapists for the Saturday Martha was raped it would tell us the probability of each of the rapists being at the gift shop at that time.”
“Yes,” said Sam. “When I got to that point my nephew said the result is likely to be useless because it’ll have a lot of low probabilities. More information is needed to narrow the results.”
“Did Jim tell you how we can narrow the results?” asks Mary.
“He did. He said if we input all the available data on Martha it might help if the rapist targeted Martha.”
“How much confidence does Jim have this AI-based software will point us to Martha’s rapist?” asks Pam.
“I asked him, Pam. He said he doesn't know. The algorithm is new. Testing has only started. They are still adding new things. Jim is currently working on how to add the cell phone contact tracing algorithms developed during the Covid-19 pandemic. The part using data on known criminals is close to being verified. It’s harder to verify for cases with specific victims because of privacy issues. He said our case is a perfect test case. That is if we can get Martha’s parents to agree to let us use all of the available data relating to Martha. Even then, we will have to enter the data on a network that’s not public. This new AI-based approach is promising; the University doesn’t want to make a mistake killing the research before it’s proven and privacy issues are resolved.”
“We can use the State’s network for criminal data. It’s not public and I can access it from my computer in the Courthouse,” says Mary.
Peter says, “I know the Crawfords well. If you’ll write out what’s needed for their consent to use Martha’s data, Sam, I’m sure I can get Martha’s parents to sign it. And I’ll caution them not to tell anyone. We don’t want the fear to return if people find out we are still looking for Martha’s rapist.”
Mary asks, “Pam if we do this will you guide us?”
“Yes, I’m excited to see what this new criminal justice software can do. I keep reading that artificial intelligence is going to change nearly everything. If it makes catching criminals easier that’s wonderful.”
Sam says, “Good. I think we have a plan. I’ll write up a statement of consent to use Martha’s private data and drop it off at your business tomorrow, Peter. Then I’ll contact my nephew, Jim. He’ll tell me what Mary has to do to access the AI-based software at the Dearborn campus. Better yet, I’ll put him in touch with you, Mary. That way you two computer experts can work it out without me screwing it up. By the time we meet next week, Peter will have had time to get the Crawfords to sign the consent form and Mary will know how to work with the AI algorithm.”
Pam says, “And I’ll think about the case. There still may be ways for old fashioned police work to be useful. I think we did well tonight. Let’s go home. Joe, will you lock up?”
“Glad to, Pam.”
No comments:
Post a Comment