AI can detect 90% of crimes before they happen

Question

An article by Jack Dunhill a Social Media Coordinator and Staff Writer for IFL Science, posted an article claiming that a "new" type of AI was able to detect 90% of crimes before they happen. The title of the article and claim happen to be misleading, essentially the AI doesn't detect crimes before they happen but uses data from past crimes in a 1,000-square-foot area of Chicago to predict when and where crimes are more likely to occur based on previous criminal data. Additionally, the "90% accuracy" they mention refers only to how well the AI predicts general crime hotspots and not exact events. The article also leaves out the important limitation of the fact that not all crimes are reported, which can affect the accuracy of its 90% prediction rate. So therefore, this AI doesn't detect crimes before they happen but instead predicts the likelihood of crimes occurring, giving police and other first responders a head start in responding to potential crime rather than stopping specific crimes before they occur.

Answer 1

The claim that AI can detect 90% of crimes before they happen is misleading and exaggerated. According to the source from IFL Science, the AI technology relies on an event log of past crimes, including where and when they were committed. The technology from this algorithm then uses the time series to generate where the next predicted crime will happen. This technology is not actually predicting specific crimes it is just predicting what is most likely to happen based on past events.

While researching, I found an article published by the University of Chicago that explains this claim closely. Just like the IFL Science article, it mentions that the system relies on pattern recognition and that the forecasted crime is not a guarantee that it will actually take place. However, this article also brings in a new perspective on bias and highlights that law enforcement and police officers are influenced by socioeconomic context, not just crime levels. The evidence shows that in wealthier neighborhoods, there tend to be more arrests as opposed to poor neighborhoods, which have fewer arrests even when crime occurs. The new AI system is revealing bias in response, not just in the data.

After comparing these sources, I can conclude that although this model helps predict when crimes are statistically likely to occur, the original title is misleading. 

Sources:

https://biologicalsciences.uchicago.edu/news/algorithm-predicts-crime-police-bias

https://www.iflscience.com/ai-predicts-90-percent-of-crime-before-it-happens-creator-argues-it-wont-be-misused-65025

Answer 2

The claim that AI can detect 90% of crimes before they happen is misleading. After reviewing the available information, I found that the AI being discussed does not actually predict specific future crimes. Instead, it analyzes historical crime data from a limited area in Chicago to identify locations where crime is statistically more likely to occur. The “90% accuracy” figure refers only to how well the model predicts general crime hotspots, not actual individual criminal events. Secondary sources such as MIT Technology Review and the Brennan Center for Justice explain that predictive policing tools are based on past crime data and often reflect biases in reporting and policing, which can reduce accuracy and fairness. The main evidence supporting the claim is that these systems can sometimes correctly identify high risk areas, but the evidence against it is stronger because the technology cannot truly detect or prevent specific crimes before they happen. Attempts to contact the original source did not provide additional clarification beyond what was already included in the IFLScience article.

Comments

Please include URL sources links for the secondary sources you mentioned (MIT and BCJ. Thanks!

Answer 3

this article is misleading. the article doesn't state that AI is detecting "90% of crimes" before they happen, but is instead just taking in data about past crimes in chicago neighborhoods and using an algorithm to analyze the data to determine potential crime hotspots based on past history. the system just estimates a probability of certain crimes happening in certain areas and in no way predicts "90% of crimes" before they occur.

Comments

What source did you use for your fact-check? It's always best to provide additional sources besides the one posted by the OP. Thanks!

Answer 4

this article is misleading. the article doesn't state that AI is detecting "90% of crimes" before they happen, but is instead just taking in data about past crimes in chicago neighborhoods and using an algorithm to analyze the data to determine potential crime hotspots based on past history. the system just estimates a probability of certain crimes happening in certain areas and in no way predicts "90% of crimes" before they occur.

Answer 5

What I found about this is it makes a claim that AI can detect 90% of crime before it happens.
2. I see that the research behind this claim is to predict this policy and see if it actually works and has been tested out in many cities.
3. The IFS is a secondary source because it reporting on a study that they are doing.
4. The biases this might have is promoting technology and saying that it is the best way to detect crimes before it occurs. Also the attention grabbing headlines and trying to be more entertaining rather than highlighting.
5. What supports the claim is not everything can be predictable so how can you know something before it happens especially when it comes to crimes and how is it that easy to measure?
6. What undermines the claim is that you can't always detect crime and it is not easy to do so, so it is not valid.
7. Overall this source is very misleading with information that is not always true.

Answer 6

 The AI does not identify specific criminals or predict individual, unique events like in the movie Minority Report.

Instead, it uses historical data to identify geographic hotspots and time patterns.

• What it actually predicts: The model divides a city into 1,000-square-foot "tiles" and predicts the probability that a crime (like theft or assault) will occur in that specific square within a one-week timeframe.

• University of Chicago News: The official university report explains that the algorithm forecasts crime patterns a week in advance but also highlights how the data reveals systemic biases in police response between wealthy and disadvantaged neighborhoods.

• BBC Science Focus: An interview with the study’s creator where he explicitly clarifies that the AI cannot be used to put people in jail before they commit crimes, as it has no capability to identify individuals.

•  

Answer 7

The original claim comes from a 2022 University of Chicago study in which their AI model uses public data on property and violent crimes to predict incidents. The model had roughly a 90% success rate in their study, however this wasn't necessarily the point of the study. The bulk of the study focuses on the systemic biases and disparities in crime enforcement, with the AI model being a tool to find imbalances in the system rather than a surveillance tool. The model also doesn't predict crimes specific crimes, rather it uses the data sets to predict where something might happen. The claim that AI can predict 90% of crimes before they happen is not necessarily incorrect; however, it requires context.

The following link goes to the original 2022 study. 

Event-level Prediction of Urban Crime Reveals Signature of Enforcement Bias in U.S. Cities

The following link goes to a 2022 article written by the University of Chicago summarizing the findings from the study.

Algorithm predicts crime a week in advance, but reveals bias in police response