1. Legal and Ethical Issues
The first thing that came to mind, before even delving into the article; and solely based on the definition provided – is how this tactic of predictive policing sounds like bias or discrimination. The key word of “target” just does not sound right, I feel like there is a right way of doing this and there’s a wrong way. In my opinion, if there’s a know location where crime is constantly being committed then yeah we probably should watch that location and try to solve the issue. I think issues arise when higher police presence in an area leads to people who are innocent and just going about their business to be questioned or harassed. The example in the text of “For example, if predictive policing software shows that a bar sees heightened crime at 2 a.m. on Saturday nights, a police department might deploy more officers there” makes sense right away but at the same time what if it was me walking down the street at 2 a.m. – Which I am perfectly allowed to do but I get harassed just because it’s in the vicinity of where others may have committed a crime. This will only contribute to bias, transparency issues, and a negative effect on public relations.
Another issue that comes to my mind is a lack of resources or other areas suffering because one place needs to be focused on, a crime can happen anywhere at anytime. If we are allocating extra resources in every place that theirs a high crime rate what about the places that may only get it every once and a while, do they just need to wait longer because we are focusing in on one town or neighborhood. I feel like this just makes people uncomfortable, and at the end of the day safety and being able to help people where they need it and when they need it should be our #1 priority.
Applying what we’ve learned
Ultimately, I think the theory behind predictive policing makes sense – we just have to find a balanced way to go about it and one that makes sense. When it comes to data mining and machine learning, I think they can be tools that are applied to scenarios like this but at the end of the day you are talking about something that’s going to have an effect on the people around you. This has to be approached with fairness, transparency, and the respect of the people who you are going to be policing – keeping in mind that most of the people aren’t going to be the ones causing any issues and don’t want to be harassed or under a microscope.
Why use a logistic tree
I believe a logistic tree would be helpful in a situation like this compared to a neural network because of its simplicity and ease of understanding and implementation. This is something that almost anyone could take a look at and get a mild understanding – people want to know what’s going on and be able to understand the method behind the madness. A neural network for something like this takes a TON of data and can easily be overly complex or misunderstood, I think their is enough misunderstanding about this whole situation to begin with so maybe it should be more simple and logical.
Something similar
Another situation like this, that could potentially be solved with data mining but could cause ethical considerations is the a hiring or application process. I think this is very similar to the predictive policing and raises similar issues, it just doesn’t really make me feel right because it feels like we are judging a book by its cover. Just looking at values or statistics on a paper and not knowing what’s on the other side, who the person is you are making assumptions about based on their experience, age, where they are from.
2. The boycott by mathematicians against predictive policing software shows serious worries about how ethical and how properly mathematical formulas are used by the police. Here are some main concerns and ideas related to this issue:
Ethical and Legal Concerns:
Bias and Discrimination: Algorithms might keep adding to old biases if the data they learn from already has prejudices. This could lead to unfair actions like targeting minority groups more than others.
Transparency: It’s often not clear how these algorithms work or what data they use. This lack of openness can stop the public from being able to oversee and trust these systems.
Consent and Privacy: Using these algorithms brings up issues about whether people agree to be watched and how their privacy is affected by a lot of data gathering.
Accountability: When mistakes happen, it’s hard to figure out who is responsible, especially when the process behind decisions isn’t clear.
Applying What’s Learned in School to Real Problems:
Discussing predictive policing in schools can show how important it is to use math responsibly and think about its effects on society. This discussion emphasizes the need for teamwork involving ethicists, community members, and legal experts when creating and using such technology.
Why Use Simpler Models Like Logistic Regression or Decision Trees Instead of Complex Neural Networks:
Interpretability: Logistic regression and decision trees make it easier to see how inputs affect outcomes, which is important for checking and understanding decisions made by algorithms in policing.
Simplicity: These models are less complex, need less data to work well, and make it easier to spot and understand mistakes.
Transparency: These models let people more easily check and understand the steps in decision-making, which is crucial in situations where you need to explain or defend decisions in public or in court.
Similar Ethical Issues in Another Area: Loan Approvals:
Just like predictive policing, loan approval algorithms use past data and can keep unfair biases alive. They might unfairly judge people based on social or economic backgrounds. Solving these ethical problems also requires thinking about bias, fairness, transparency, and who is responsible.
