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

This claim has been proven to be true by some sources, but it takes a lot for this to be true. Finding that AI can detect crime before it happens have been conducted and proved effective but it is still difficult to tell the future, even for AI.

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

Answer 2

“AI can detect 90% of crime before it happens" is a very misleading claim after you look at the original source published by IFL Science. According to the article, AI can be used to "predict crimes before they happen". However, looking into the research cited in the paper and was published in Nature Human Behaviour, you can understand that the AI model predicts patterns based on data from previous crimes committed in a specific geographic area. 

The University of Chicago describes how the algorithm functions. It is based on information about the time and place of a crime's occurrence, you can use it to estimate that some places will have some crimes committed in the near future.

You can find more information regarding the limitations of the algorithm in the original article as well. For example, it is mentioned that "AI doesn't tell you who is going to commit the event". The algorithm can only show what is likely to happen in a very specific location.

As a whole, the statement is incorrect. It should be remembered that the AI system does not predict specific crimes at all, it only detects that there is a pattern of crime in a location, and can only be done with statistical terms. Regarding the “90% accuracy,” it exists, but only in relation to pattern detection.

https://www.nature.com/articles/s41562-022-01372-0?utm

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

Answer 3

I think this article is really kinda misleading. It talks about AI being used in multiple US cities and says it has over 80% accuracy in predicting crimes, but that’s not really the full story. It’s not actually a new AI it is just using an algorithm that looks at past crime data in specific high crime areas to predict the chances of something happening there again. The problem is it is only focused on small areas where crime is already common, so of course the accuracy looks high, but that doesn’t mean it works the same everywhere.

Answer 4

This claim is overall misleading as AI does not actually detect crimes before they happen, but instead predicts where crimes are more likely based on data. The IFLScience article https://www.iflscience.com/ai-predicts-90-percent-of-crime-before-it-happens-creator-argues-it-wont-be-misused-6502 makes it sound like AI can stop crimes before they happen, but it actually just uses past crime data to predict where crime is more likely. The “90% accuracy” only refers to guessing general areas, not specific crimes. This means the AI cannot truly predict the future, only patterns. The article may be biased because the title is very attention-grabbing and exaggerates the results. The evidence supporting the claim is that the AI can find patterns in crime locations. The evidence against it is that not all crimes are reported and it cannot predict exact events. Overall, the AI helps police focus on higher-risk areas, but it does not detect crimes before they happen.

Answer 5

In investigation of the topic that a new AI has been made that can detect 90% of crimes before they even happen, I found that the claim was in fact correct. After reading articles by the BBC and also IFL science. 

In the BBC article that I read, I learned that “The model doesn’t predict who will commit a crime, but rather identifies patterns in where and when crimes are likely to occur". This article supports the idea that newer AI systems can predict crime patterns with up to 90% accuracy by analyzing past data, showing how technology is shifting toward anticipating crime before it happens rather than just reacting to it.

In the IFL science article that I read, I learned that researchers developed an AI system that analyzes crime data to predict where and when crimes are likely to happen up to a week in advance with about 90% accuracy, showing how AI can be used to anticipate crime patterns before they occur rather than only responding after the fact. 

Predictive crime AI can be biased because it learns from past police data, which may already focus more on certain neighborhoods, making them seem more “criminal” than they really are. This can create a cycle where those areas keep getting more police attention, which keeps reinforcing the same prediction.

Answer 6

In investigation of the topic that a new AI has been made that can detect 90% of crimes before they even happen, I found that the claim was in fact correct. After reading articles by the BBC and also IFL science I learned that it may be a bit exagerrated though.

In the article written by the BBC, the following statement was made: "The model doesn’t predict who will commit a crime, but rather identifies patterns in where and when crimes are likely to occur". This article clearly shows that new AI models predict crime patterns up to 90% accurately. This proves that technology now tends to predict and anticipate crimes based on analysis of previous events.

In the article published on IFL Science website, the authors state the following: "Researchers have designed an AI-based model capable of predicting when and where crimes might take place up to one week in advance with 90% accuracy". Thus, we see that new AI systems have become able to anticipate crime patterns.

Bias is quite possible in case of AI for predicting crimes because the algorithm of such AI is based on the analysis of previous crime patterns. As police tend to patrol certain places due to the increased amount of crime incidents, certain areas can look "criminal" in reality, and AI models can continue learning from those predictions.

Answer 7

No AI or artificial intelligence nor human can detect crime before it happens. It is not possible to look directly into the future and predict exactly what is going to happen. However, the 90% number here is really referring to an AI model developed by the University of Chicago that analyzes part crime data and can predict where crime is most likely to occur in the near future. 

Researchers at the University of Chicago created an AI model that directly predicts crime hotspots, which is not the same as being able to predict what crimes are going to happen. This means rather it can pinpoint the area that crime is most likely to occur in say for example a city, and can pinpoint which areas within that city need to be watched more thoroughly. The model cannot identify people or specific events. Researches also warn that the system exposes policing bias, not who is going to be the next criminal (Lerner). 

The Guardian has an article that goes more into depth on the matter. It states that the AI analyzes crimes in major cities such as Chicago, Atlanta, and Los Angeles. It divides the cities into small geographical squares that are about 1000 feet across, which then are used to see where certain types of crime are going to occur the following week. This is exactly where the 90% accuracy comes from when referring to predicting crime hotspots, because it is not predicting who or what crime is going to happen, instead just the area. Critics also note that is it very important to understand the model is already biased due to reporting crimes that have already occurred. It also reveals policing disparities, such as wealthier neighborhoods, they had fewer arrests despite having similar crime patterns. This predictive policing rewards racial and socioeconomic biases, thus they should be used extremely cautiously (Helmore). 

Thus, this fact cannot be proven to be correct. It is not possible to use AI to predict what is going to happen before it actually has occurred, even though research can be used to show where crime may be more likely to happen due to hotspots based on past data. However, it is so important that moving forward we are aware of the consequences of using data like this. Because often times, reported crime is most likely to come from lower income areas and can racially target those who are within these locations. These predictions come with limitations, biases and most importantly ethical concerns that we need to continue to be aware of with AI becoming such a big part of our lives today. 

Lerner, Louise. “Predictive Policing Model Shows Promise, Reveals Bias in Police Response.” University of Chicago News, 30 June 2022, news.uchicago.edu/story/predictive-policing-model-shows-promise-reveals-bias-police-response.  

Helmore, Edward. “AI Can Predict Crime in US Cities a Week before It Happens, Researchers Say.” The Guardian, 1 July 2022, www.theguardian.com/us-news/2022/jul/01/ai-predict-crime-us-cities-chicago