Donna Napolitano
Forensics, E/F Block Odd
Current Event
9-27-21
Article Link:
Matthew Robert Bennett Professor of Environmental and Geographical Sciences, and
Marcin Budka Professor of Data Science. “We Trained Ai to Recognise Footprints,
but It Won't Replace Forensic Experts Yet.” The Conversation, 18 Aug. 2021,
https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replac
-forensic-experts-yet-161686.
Matthew Robert Bennett’s article explains how and why forensic scientists have trained an AI to detect and analyze footprints from a crime scene. As a result of bloody footprints being an extremely common sight and piece of evidence at crime scenes, investigators typically analyze these prints in an attempt to reconstruct events and to sometimes profile a suspect. Additionally, as stated above, shoe prints are one of the most common types of evidence as their traces are typically found near window sills, doors, floors, etc...and the size of these footprints are often used to figure out a suspect’s height, weight, and even gender. In a recent study, a group of scientists created a neural network (which is a form of AI) and asked the AI to analyze a set of footprints and to determine the person’s gender through its analysis. The AI guessed the gender of the person with 90% accuracy. Furthermore, because analyzing footprints is a tedious process, experts created the AI in an attempt to increase the process’s accuracy. For instance, in one experiment, a footwear investigator was given 100 randomly selected shoe prints to analyze, yet the investigator’s accuracy only ranged from 22% and 83%. While the AI was given the same set of data and had an accuracy ranging from 60% and 91%, therefore the AI seemed to be a good substitute for a footprint specialist and, in the end, can help to increase an analyst’s accuracy. All in all, the AI had been created to aid forensic scientists in reducing the amount of work they need to do when solving a case, helping them to focus on other aspects of the crime while the AI can be used to analyze footprints.
As a result of shoe prints being one of the most common types of evidence found at crime scenes, the creation of this new AI that can help investigators analyze footprints in a more accurate manner. This AI seems to be the type of technology that could take the burden off of investigators’ chests, aiding them in focusing on more complex cases, rather than the analysis of crime scene footprints. For instance, because the size of footprints can help investigators figure out a suspect’s height, weight, and even gender, the AI can be used to do the same, while investigators can analyze different aspects of a case, helping to move it along quicker than before. To add to that, when comparing an AI and a footprint investigator, the AI’s footprint analysis accuracy was higher than that of the investigator, proving how important the AI can truly be throughout forensic investigations and why this method needs to be utilized across the globe. Furthermore, the article also mentioned how there are fewer than thirty footwear experts in the UK today, therefore the AI can also serve as a replacement for these types of experts and help to increase their productivity. Overall, because analyzing footwear is fairly challenging, the work of these investigators often needs to be looked over by another expert and the AI can be the one to do so.
After reading Bennett's article, I noted a variety of strengths and weaknesses. First off, I enjoyed how he started off the article with examples of why footprints are common to see at crime scenes and why they are so crucial to the investigation as his examples and explanations aided readers in truly understanding the importance of footprint analysis and why creating an AI to analyze these prints would be so useful. I also enjoyed how Bennett was able to back all of his claims with evidence and explain the importance of footprint investigators when working on a case. However, I did feel as if his article was a bit repetitive at times as he seemed to state the same few things about the importance of footprints and why the AI had been made in a variety of ways, instead of bringing up new points. Basically, I wish Bennett had altered the wording he had utilized in the article, adding a few more interesting details in order to make his writing more engaging and maintain a range of things he was talking about.
7 comments:
Eglis Gjonpalaj
Article link:
Matthew Robert Bennett Professor of Environmental and Geographical Sciences, and
Marcin Budka Professor of Data Science. “We Trained Ai to Recognise Footprints,
but It Won't Replace Forensic Experts Yet.” The Conversation, 18 Aug. 2021,
https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replac
-forensic-experts-yet-161686.
In the article Donna wrote, I really liked how she summarized the article. It made me understand all the points she was trying to make. Another thing I liked was that she did a great job showing how shoe prints are important today and how it affects many parts of Forensics science. Her thoughts on the article were very interesting, this was because she pointed out the strengths and weaknesses the article showed.
One thing she could have improved on was the summary and how precise it was. She could have done a better job of making sure the reader understood what was going on. Another thing she could have improved on is her analysis and how it could have been more clear.
One thing that shocked me was how important footprints still are today. I assumed since technology is so advanced we were able to look past footprints, but it is quite the opposite.
Derek Bilyeu
Mr. Ippolitto
Three aspects that were well presented in this passage are the following; why forensic scientists have trained an AI to detect and analyze footprints from a crime scene, their traces are typically found near window sills, doors, floors, etc, and As a result of shoe prints being one of the most common types of evidence found at crime scenes, the creation of this new AI that can help investigators analyze footprints in a more accurate manner.
I do have 2 suggestions on how the article can be better. My first suggestion is that there should be more examples of why scientists trained these AI. Also I feel like there should be some more information about the neutral network and how it works. Other than that I think it was a pretty well written article. Made some good points and gave good detailed evidence.
There was one aspect that I thought was very interesting in this article. That aspect was how the forensic scientists created the AI to detect certain things on a crime scene. This is amazing how advanced technology is getting. Also the AI can probably spot things that the ordinary human can't and this can be useful and more beneficial.
Matthew Robert Bennett Professor of Environmental and Geographical Sciences, and Marcin Budka Professor of Data Science. “We Trained Ai to Recognise Footprints, but It Won't Replace Forensic Experts Yet.” The Conversation, 18 Aug. 2021, https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replac-forensic-experts-yet-161686.
https://bhscsi.blogspot.com/2021/10/we-trained-ai-to-recognise-footprints.html Won't Replace Forensic Experts Yet (bhscsi.blogspot.com)
One thing that Donna did well on her review of the article “We Trained Ai to Recognise Footprints, but It Won't Replace Forensic Experts Yet.” was the usage of a detailed explanation on the AI described in the article as it allowed the reader to create a perception on how the AI can be used in broader terms. Another thing that Donna did well was the showcasing on how the AI was a good addition to the forensic lab tools by being able to be better at analyzing shoe prints over an investigator. This helped by showing the true abilities of the AI over humans, which in turn showcase its worthiness. Finally, a third thing that Donna did well in her review was the explanation of the necessity of this AI, by explaining how in the UK there are less than 30 experts in this field, making this sort of tool easily accessible to most regions in the UK as a substitute.
One thing that Donna could have improved upon was explaining how experts are inaccurate compared to the AI, such as simple human error or if the AI is able to detect things that the experts can not. Another thing that she could have done better would have been to explain how the AI was built from it’s background such as the company or government that made the AI and what it was currently being used for.
Overall, I felt that Donna’s review was great, and it allowed me as a reader to understand in depth the benefits to use the AI in footprint analysis. I learned about the scarcity of experts which surprised me given how it would be a generally well educated field but I was mistaken. This will change my perception on forensic scientists and how they specialize their professions as it seems many critical stages are not being staffed to aquadate levels.
Derek Bilyeu
Mr. Ippolito
There were three aspects that were well presented in this article. One being; Matthew Robert Bennett’s article explains how and why forensic scientists have trained an AI to detect and analyze footprints from a crime scene. Another is; As a result of bloody footprints being an extremely common sight and piece of evidence at crime scenes, investigators typically analyze these prints in an attempt to reconstruct events and to sometimes profile a suspect. Additionally, as stated above, shoe prints are one of the most common types of evidence as their traces are typically found near window sills, doors, floors, etc… And finally’;the size of these footprints are often used to figure out a suspect’s height, weight, and even gender.
I do have some suggestions for the article though. I feel as if there needs to be more details regarding the topic in general. I also believe that there needs to be more evidence presented of tests that have taken place.
There was one very nice aspect that I took away from this article. That aspect being how and why forensic scientists have trained an AI to detect and analyze footprints from a crime scene. I was very fascinated on this and thought it was very interesting. Overall I thought this was a very strong and interesting article.
James Balseiro
Mr.Ippolito
Forensics
10/20/21
Citation
Bennett, Matthew Robert, and Marcin Budka. “We Trained AI to Recognise Footprints, but It Won't Replace Forensic Experts Yet.” The Conversation, Innovate UK, 6 Oct. 2021, https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replace-forensic-experts-yet-161686.
The article talks about an AI created to identify the type of shoe someone is wearing. When given 100 shoe prints to analyze the AI was correct 60%-901% of the time. Comparing that stat to the 100% that footwear experts got shows that the AI is not ready to be used in crime labs. The AI didn’t work well with the shoe prints due to daily wear and tear a shoe goes through. The experts were able to detect the wear and tear and adjust, but the AI wasn’t. While the AI didn’t beat the experts, it did beat the casual. The article brings up the possibility of the AI being used alongside the expert sort of as an assistant. The article ends with the statement that AI will never be able to completely replace the experts, but it might be able to lessen the burden on them.
I felt that this article was very relevant. With the growing automation of many jobs such as manufacturing, I felt that this article was very relevant. Jobs that at one point seemed untouchable such as doctors are now under the threat of being replaced by AI. If replacing human shoe experts with AI could save lives then there's no question that they should be brought into service. Maybe it will cost people some jobs, but even the article states that in the UK right now there are less than 30 shoe print experts. Less than 30 people losing their jobs for the possibility of hundreds of lives being saved is an easy decision.
A critique I have about the article was the ending. AI replacing shoe print experts seems very realistic, but the authors just write it off after one AI fails. The authors seem very quick to judge the trial as a failure, but as far as the article states the attempt is the only known one. If there was another then the article should’ve stated it and its results. A strength the article has is that it does a good job explaining what went wrong with the AI and what the humans did better. Learning that the AI had a hard time adapting to wear and tear was super interesting to read about.
I thought that Donna did a great job at explaining firstly, how footprints are important in forensics science and in criminal/civil cases. Secondly, I thought that Donna did a great job at expanding on how artificial intelligence is being used in the forensic science field, especially because it is becoming a much more advanced and (hopefully) efficient use of technology. Thirdly, I thought that Donna did a good job on explaining why artificial intelligence may not be ready to use yet: specifically, I like how she explains the fact that shoe size varies and can wear down, making it difficult to track through AI.
I think that Donna could have elaborated more on the success of AI use in the forensic science field (if the article spoke about it), and how (if it is) currently being used (in more detail. What are the success rates?). Secondly, I think Donna could have spoke more about what it means if AI is a groundbreaking advancement in technology that can be used effectively (what impact does it have on forensic science?),
I learnt a lot about this review, and knew that AI was being used in many fields, but not directly for footprint identification. I was surprised that in the report, AI did outperform many other users. I think it’s an important topic because as mentioned in the review and article, “ it could help free up time for the expert to focus on more difficult cases”.
Sofie Blazejczak
Oct 20, 2:45 PM
Julia Bazinet
Mr. Ippolito
Forensics / Current Event
10/18/21
Article Link: https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replace-forensic-experts-yet-161686
Matthew Robert Bennett Professor of Environmental and Geographical Sciences, and Marcin Budka Professor of Data Science. “We Trained AI to Recognise Footprints, but It Won't Replace Forensic Experts Yet.” The Conversation, 6 Oct. 2021, https://theconversation.com/we-trained-ai-to-recognise-footprints-but-it-wont-replace-forensic-experts-yet-161686.
I really enjoyed reading Dona’s review on her article. I thought she did a wonderful job of introducing the article. She started off with a vague topic sentence, and then went into detail about why the article is relevant. This was a great way to introduce the article, many times reviews of articles just stay the title and author of the article and explain what the author said. Dona’s way was more intriguing and effective. I also liked how she gave background on footprints. For example, she said that “shoe prints are one of the most common types of evidence as their traces are typically found near window sills, doors, floors, etc...and the size of these footprints are often used to figure out a suspect’s height, weight, and even gender.” This was one of her details that she gave about shoe footprints, and I found it helpful. Without deilas like these, I would understand why scistits would have developed AI to detect and analyze footprints. Lastly, I thought it was great that she starstics. Stats are always helpful when trying to understand what the author is talking about. The specific statistic she used that I liked was when she talked about the accuracy of the AI footprint detector: “While the AI was given the same set of data and had an accuracy ranging from 60% and 91%, therefore the AI seemed to be a good substitute for a footprint specialist and, in the end, can help to increase an analyst’s accuracy.”
While I thought Dona did a great job revving this article, I thought there were a few things she could approve upon. First, is her organization. I think her review would improve a lot if she divided it up into more paragraphs. The paragraphs that she has are too long, and a bit overwhelming. Second, if Dona compared the process of the AI footprint detector to the process of an actual person detecting a footprint, it would be good. This comparison would strengthen her article by showing how great this invention is. So, maybe she could make a new paragraph that just focuses on comparing the two.
This article reminds me of how technology is helping the forensic science field a lot. Further, as technology gets more advanced, crime cases are going to get easier to solve. Personally, I chose Dona’s article because AI is something I am intrigued by. I will now be more aware of how AI can affect forensics. More so, as technology advances I will consider how it affects AI.
Post a Comment