AI and Facial Recognition in Crime by Kai Kim-Suzuki

25khkimsuzuki — 2023-10-24 13:02:56 UTC

SOLVING CRIMES (FBI)

25khkimsuzuki — 2023-10-25 14:37:14 UTC

The FBI uses facial recognition algorithms to search for suspects of crime. This software is provided by third parties. This software is then used by the FACE services unit, which compares the faces of suspects to the faces from government databases. FACE stands for Facial Analysis, Comparison, and Evaluation. The unit then filters through potential matches manually to quickly narrow down the search for the identity of the suspect. According to the FBI, this process only provides "most likely candidates", and not positive identification.

The Unit has performed 390,186 searches on different databases to support FBI investigations.

25khkimsuzuki — 2023-10-25 14:37:19 UTC

25khkimsuzuki — 2023-10-25 14:47:13 UTC

PRIVACY

25khkimsuzuki — 2023-10-25 15:10:24 UTC

Although the FBI claims they only use "government data" such as mug shots, they fail to mention their full access to driver's license photos in 16 states -- giving the FBI access to hundreds of thousands of adults' faces in addition to their 30 million photos from a biometric database. The issue is that the government does not notify the population about this. From the Libertarian ethical perspective, this is a massive invasion of personal freedoms. Unless every person who has their face in the government database has agreed to it, the government has no right to keep track of them, as that is not the purpose of a government from the libertarian standpoint.

FALSE POSITIVES & RACIAL BIAS

25khkimsuzuki — 2023-10-27 02:15:25 UTC

The F.B.I. has no system to find how often the facial recognition algorithm identifies a false positive. There is even a concern that the algorithm, depending on how it is trained, can have a racial bias, identifying white subjects more accurately than black subjects. In a country in which African-American incarceration rates are much higher than other racial groups, this issue can be extremely problematic. Porcha Woodruff, a Black pregnant mother, was arrested this year for robbery in Detroit. It was the first known false positive arrest due to facial recognition. From the Bentham Utilitarian perspective, this is OK, because only a small minority will be negatively affected, and the majority will benefit from reduced crime.

25khkimsuzuki — 2023-10-27 02:15:48 UTC

https://www.theatlantic.com/technology/archive/2016/04/the-underlying-bias-of-facial-recognition-systems/476991/

RAPID CRIME RESULTS

25khkimsuzuki — 2023-10-27 02:28:45 UTC

Unlike human criminal investigators, a facial recognition algorithm can parse through billions of faces in minutes. Facial recognition helped arrest the January 6th insurrectionists, with great efficiency, as there was lots of video data of the faces of participants. This aspect of speed is also critical for cases involving kidnappings, where a manual investigation might just be too slow. From a utilitarian standpoint, this tech is good, as it only does a human investigator's job, but much quicker. Therefore, there is a large increase in safety, and thus happiness.

CHINA

25khkimsuzuki — 2023-10-27 17:04:09 UTC

China in recent years has abused the power of AI and facial recognition to further the authoritarian regime's agenda. Although they claim to use it to improve society, they have gone as far as to have people shunned for minor criminal offenses on billboards using the technology. Behind the scenes, they use biometric data and behavior data along with AI to identify citizens who have the potential to threaten the state. Those who are flagged are arrested and punished. Crimes include being Muslim.

25khkimsuzuki — 2023-10-27 17:04:23 UTC

https://ideasforpeace.org/content/human-rights-in-ai-facial-recognition/

ABUSE OF POWER (CHINA)

25khkimsuzuki — 2023-10-27 17:15:10 UTC

China has been discriminating against its Uyghur minority population for a decade, subjecting Chinese Turkic Muslims to mass detention camps, torture, and execution. China uses facial recognition to racially profile Uyghurs for arrest. In a single month, the city of Sanmenxia has even run this Uyghur-identifying algorithm 500,000 times on its residents.

IMPROVED BEHAVIOR (CHINA)

25khkimsuzuki — 2023-11-03 21:23:17 UTC

China has one of the lowest crime rates and enjoys some of the best security in the world. It is doubtless that facial recognition technology has played a large role in this. China is conducting a massive behavioral engineering agenda, which uses facial recognition to humiliate J-walkers, shoplifters, and other people who have committed minor offenses. This creates a sense of fear in the citizens, strongly discouraging any rule-breaking. From a Virtue-Ethics perspective, this is not an issue. This method China is using promotes virtuous people. The question is, how is a virtuous action defined, and by whom?

25khkimsuzuki — 2023-11-03 21:27:21 UTC

25khkimsuzuki — 2023-11-03 21:27:28 UTC

COLD CASE RESOLUTION

25khkimsuzuki — 2023-11-03 21:53:34 UTC

Homeland Security has recently begun using facial recognition to restart cold cases, such as child exploitation cases. With the use of Clearview AI, Homeland Security has solved years-old cases that have brought hundreds of previously un-caught pedophiles to justice. From the Kantian perspective, this reward is positive because the intentions are very much in the right.

TRANSPARENCY POLICIES

25khkimsuzuki — 2023-11-03 22:05:24 UTC

Governments could pass laws and policies that require the FBI and Homeland Security to be transparent about their use of facial recognition, as well as the software itself. This will increase public awareness, and should an issue arise, the public can push for change.

BIAS MITIGATION

25khkimsuzuki — 2023-11-03 22:11:40 UTC

Legislature could be passed to ensure that every person with access to facial recognition technology is trained to mitigate bias. In addition, the training data used to create the facial recognition models could be ensured to be diverse, and inclusive of all races. This would dramatically reduce the number of harmful false positives and negatives.

DATA COLLECTION CONSENT

25khkimsuzuki — 2023-11-03 22:16:55 UTC

Currently, the government scrapes up billions of faces from the internet to identify criminals un-consensually. This issue could be resolved by just asking for consent for criminal investigation use when a photo is posted to the internet. Another, less controversial way to attain consent for use could be to just have a subtle warning when someone posts a photo on the internet that says that the image may be used for criminal investigation should the image be posted.

25khkimsuzuki — 2023-11-03 22:39:58 UTC