Latest / Murder Files Unsealed / AI Could Be Used to Solve Britain’s Most Notorious Cold Cases

Transcript

AI Could Be Used to Solve Britain’s Most Notorious Cold Cases · Murder Files Unsealed

• 0:00
  Speaker 1: Thousands of individuals utilizing the London Underground were monitored through
• 0:04
  AI surveillance software that analyzed their movements, behaviours and body
• 0:09
  language to determine if they were engaging in criminal activities
• 0:14
  or were in potentially dangerous situations. As revealed by new
• 0:18
  documents obtained by Wired. This machine learning technology was integrated
• 0:24
  with live CCTV footage to identify aggressive actions, the presence
• 0:29
  of weapons such as guns or knives, and incidents of
• 0:32
  individuals falling onto the tracks or evading fare payment. From
• 0:38
  October twenty twenty two to the end of September twenty
• 0:42
  twenty three, Transport for London TfL, which manages the city's
• 0:47
  tube and bus services, conducted tests on eleven algorithms to
• 0:52
  observe individuals at Wilsden Green Tube station, located in the
• 0:56
  northwest of London. This proof of concept trial marks the
• 1:00
  first instance in which the transport authority has utilized AI
• 1:05
  alongside live video feeds to create alerts for frontline personnel.
• 1:11
  During the testing period, over forty four thousand alerts were generated,
• 1:16
  with nineteen thousand of these being communicated to station staff
• 1:20
  in real time. Documents provided to Wired in response to
• 1:24
  a Freedom of Information Act request reveal that Transport for
• 1:28
  London TfL employed various computer vision algorithms to monitor individual's
• 1:34
  behavior within the station. This marks the first occasion on
• 1:38
  which comprehensive details of the trial have been disclosed. This
• 1:43
  development comes after TfL announced in December its intention to
• 1:47
  broaden the application of artificial intelligence to identify fair evasion
• 1:53
  at additional stations throughout the British capital. In the trial
• 1:57
  conducted at Wilsden Greenation that previously accommodated twenty five thousand
• 2:03
  visitors daily prior to the COVID nineteen pandemic, the AI
• 2:07
  system was implemented to identify potential safety incidents, enabling staff
• 2:13
  to assist individuals in need. Additionally, it aimed to address
• 2:17
  criminal and antisocial behavior. Three documents provided to Wired outline
• 2:23
  the application of AI models in recognizing wheelchairs, strollers vaping,
• 2:29
  unauthorized access, and individuals placing themselves at risk by approaching
• 2:34
  the edges of train platforms. The documents, which contain partial redactions,
• 2:40
  also reveal instances of errors made by the AI during
• 2:43
  the trial. For example, it incorrectly flagged children accompanying their
• 2:49
  parents through ticket barriers as potential fare evaders and struggled
• 2:53
  to distinguish between folding and non folding bicycles. Furthermore, police
• 2:59
  officers participated in the trial by displaying a machete and
• 3:03
  a firearm within the view of CCTV cameras while the
• 3:07
  station was closed, to enhance the system's capability to detect weapons.
• 3:12
  Privacy experts who have examined the documents express concerns regarding
• 3:16
  the reliability of object detection algorithms. They also highlight the
• 3:21
  ambiguity surrounding the number of individuals aware of the trial,
• 3:25
  and caution that such surveillance systems could potentially be enhanced
• 3:30
  in the future to incorporate more advanced detection technologies or
• 3:34
  facial recognition software aimed at identifying specific individuals. Although this
• 3:40
  trial did not utilize facial recognition, the application of AI
• 3:45
  in public settings to monitor behaviours, assess body language, and
• 3:49
  deduce protected characteristics presents numerous scientific, ethical, legal, and societal
• 3:55
  dilemmas similar to those posed by facial recognition technology, states
• 4:01
  Michael Burtwhistle, Associate director at the independent research organization the
• 4:07
  Ada Lovelace Institute. Some of the most infamous cold cases
• 4:11
  in Britain may find resolution through the application of artificial intelligence.
• 4:16
  Following a trial that demonstrated its capability to analyze evidence
• 4:21
  that would require eighty one years of human labour in
• 4:24
  under two days, law enforcement leaders are optimistic that this
• 4:29
  technology could be utilized in the future to efficiently process
• 4:34
  numerous case files, thereby assisting detectives in uncovering new investigative avenues.
• 4:41
  A trial conducted by Avon and Somerset Police employed an
• 4:46
  AI tool known as Souza, developed in Australia to evaluate
• 4:50
  twenty seven intricate cases using this digital investigation platform. The
• 4:56
  findings revealed that the AI could assess all evidential materials
• 5:01
  within thirty hours, a task that would have taken a
• 5:04
  human up to eighty one years to complete manually. The
• 5:08
  police force is now preparing to conduct a more extensive
• 5:11
  trial to further explore the advantages of this technology and
• 5:16
  adapt it for evidential purposes within UK law enforcement. Gavin Stevens,
• 5:23
  Chairman of the National Police Chief's Council, expressed that this
• 5:26
  technology could prove to be extremely beneficial in addressing the
• 5:30
  protracted reviews of cold cases that remain unsolved. The platform's
• 5:36
  website indicates that the tool is capable of efficiently analyzing
• 5:40
  various types of data, including video footage, financial transactions, call records,
• 5:46
  social media interactions, emails, images, documents, as well as information
• 5:52
  from phones and computer hard drives.