0% found this document useful (0 votes)
11 views9 pages

(7-15) Ai Policing in Criminal Justice Methods & Concerns in Crime Detection and Prevention in India

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views9 pages

(7-15) Ai Policing in Criminal Justice Methods & Concerns in Crime Detection and Prevention in India

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

JLAI

Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661

AI POLICING IN CRIMINAL JUSTICE:


METHODS & CONCERNS IN CRIME
DETECTION AND PREVENTION IN INDIA
Manoj Meena*
Aishvarya Joshi**
Abstract: Artificial intelligence no longer remains a concept to be speculated about, in 2018, Daniel
Flagella at INTERPOL in UN global meet mentioned the term AI Policing, which is a derivative theorem
of AI and Law. This legal tech tool is and can be used to catch criminals and their unlawful acts through
video technology, robotics, surveying crime scenes, facial recognition, analysis of text-based intelligence,
video and audio analysis, vehicle identification, and surveillance which can enhance the propensity to deal
with many crimes. AI algorithms through predictive policing can easily reduce the risk through the massive
amount of data on criminal activities. However, the lack of human oversight in automated processes,
inaccuracy of data, and biases in AI prediction due to social inequalities in the country are the imminent
threat of misuse, inaccuracy of judgments, and threat to fundamental rights. AI Policing suffers from
ethical and legal issues unless regulated and addressed effectively. This doctrinal study explores the
applications of AI in policing to detect and prevent the crime before its occurrence. Further, it attempts to
highlight the ethical and legal challenges in the execution of AI Policing in India. Finally, the study
suggests measures to improve accuracy and efficiency in addressing the ethical and legal issues for a smart
and robust criminal justice system.

Keywords – Artificial Intelligence, Criminal justice system, crime detection, predictive policing

1. INTRODUCTION solve their issues using algorithms. At this


A generation of mathematicians, physicists, an point, computers could store large amounts of
d philosophers became familiar with the idea of data and were quicker, more widely available,
artificial intelligence in the 1950s. By asking and less expensive. AI gained much huge
the question, “If a human is using available publicity when the top chess masters of the
data or pieces of information to find out the world were defeated by the IBM’s AI based
causes of problems, then why cannot we chess program ‘Deep Blue’ in 2011. It was
employ machines for a similar function?” Alan followed by development of speech recognition
Turing, a British polymath, came up with a softwares, humanoid robots which could
mathematical possibility of artificial depict and recognizing emotions, to AI news
intelligence. He discussed this framework in readers and AI policing. According to
his paper, “Computing Machine Learning an UNESCO, “there have been spectacular
Intelligence,” published in 1950, in which he advances in the field of artificial intelligence
discussed how machines can function (AI) in recent years, leading to inventions that
intelligently. But due to a lack of computing we had never thought possible”.2
features, he couldn’t reach an effective
solution, where computers can’t remember the The United Nations Development Initiative
tasks done and could only get the function of and the Indian government launched the
doing new tasks plus it was an expensive Knowledge-Based Computing Systems
affair.1 (KBCS) programme in 1986, which marked
the official start of AI research in India. The
Later, as AI grew, machine learning algorithms project on Machine Translation for Indian
improved, giving individuals a better means to Languages by IIT Kanpur, the project on
Optical Character Recognition by ISI Kolkata,
the flight-scheduling expert system, Sarani,
* Advocate, Bar Council of Rajasthan
**Assistant Professor, Department of Law, University of
2
Rajasthan, Jaipur ‘Towards a global code of ethics for artificial intelligence
1
AI, ‘A brief History of Artificial Intelligence’, < research’, The UNESCO Courier, July-September 2018, p.
https://indiaai.gov.in/article/a-brief-history-of-artificial- 3 <https://unesdoc.unesco.org/ark:/48223/pf0000265211>
intelligence> accessed 23 May 2022 accessed 22 May 2022

7 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
developed by CDAC, Mumbai, a speech intelligence and emulate the human capabilities
synthesis system created by TIFR for the of sense, comprehend and act”.7
Indian Railways, and an image-processing
facility using AI and vision techniques In other words, AI refers to a machine's or
developed by IISc are some of the earliest human-controlled robot's capacity to carry out
projects on artificial intelligence in India.3 tasks in the same way as intelligent humans do.
Even Nevertheless, some of AI's potential uses This phrase is widely used when building
raise ethical concerns, such as data collecting systems that have human-like intellectual
that invades privacy, face recognition abilities including generalisation, meaning
technology that may be biased against certain discovery, learning from experience, and the
races or that is intended to identify aggressive capacity to provide justifications.
behaviour. The ethical issues that AI brings are
numerous and will definitely persist in the 1.2 Predictive and Preventive Policing in
future, becoming more serious. That’s how India
computers surpassed our needs for AI through In order to identify potential targets for crime
their fast-speeding processor and heavy storage prevention or to solve prior crimes through
as estimated by Moore’s Law. police involvement, predictive policing applies
analytical tools, notably quantitative
1.1 What is AI? approaches. Even twenty years ago, before big
According to the report by AI standardisation data, law enforcement agencies like the New
committee, “there is no uniform and globally York Police Department (NYPD) were already
accepted definition of AI”. 4 Alan Turing heavily utilising crime data to track trends with
believed that “if a machine can talk to a person other indicative issues, such as the locations of
and will be mistaken to be human, then the crime victims and gun arrests. However, what
machine is intelligent”. Several definitions can is new is the amount of data available in the
be considered for understanding the notion of collection as well as the ease with which
AI including: organisations can analyse and draw
• John McCarthy defines AI as “a conclusions. Specifically, when emerging
science and engineering of making intelligent technologies allow us various applications and
machines, especially intelligent computer more rigorous interrogation of data, including
programs”.5 greater accuracy in predicting future incidences
• Kaplana and Haenleinb defined AI as of crime.8
“a system’s ability to correctly interpret
external data, to learn from such data, and to In January 2020, the Indian state, Himachal
use those learnings to achieve specific goals and Pradesh installed thousands of CCTVs to have
tasks through flexible adaptation”.6 ‘CCTV surveillance’. Jharkhand developed its
• NITI Aayog defines AI as “a IT infrastructure after getting funds from the
constellation of technologies that enable Ministry of Home Affairs. The Jharkhand
machines to act with higher levels of police utilised this 18.5 crore money for
predictive policing in association with the open
group on e-governance and the National
3
‘Artificial Intelligence in India – A Sneak Peek’, Informatics Centre (NIC).9 Similarly, many
<https://Inurture.Co.In/Artificial-Intelligence-In-India-A- other state governments have taken initiative,
Sneak-Peek/> accessed on 20 May 2022 particularly in law enforcement to use
4
‘Indian Intelligence Artificial Stack’, 2020, AI
Standardisation Committee, Department of analytical techniques to prevent or solve
Telecommunications crimes. Crime mapping, analytics, and
<https://www.tec.gov.in/pdf/Whatsnew/ARTIFICIAL%2 prediction system (CMAPS), a programme that
0INTELLIGENCE%20-%20INDIAN%20STACK.pdf> can access real-time data, was first used by
accessed on 20 May 2022
5
John McCarthy, ‘What Is Artificial Intelligence?’, Delhi police in 2015 to implement predictive
<https://homes.di.unimi.it/borghese/Teaching/AdvancedIn policing in the city. In addition, other states
telligentSystems/Old/IntelligentSystems_2008_2009/Old/
IntelligentSystems_2005_2006/Documents/Symbolic/04_
7
McCarthy_whatisai.pdf> accessed on 20 May 2022 ‘National Strategy for Artificial Intelligence’, NITI
6
Andreas Kaplana & Michael Haenleinb, ‘Siri, Siri, in my Aayog, 2018,
hand: Who’s the fairest in the land? On the <http://www.niti.gov.in/writereaddata/files/document_pub
interpretations, illustrations, and implications of artificial lication/NationalStrategy-for-AI-Discussion-Paper.pdf>
intelligence’, Business Horizons, Volume 62, Issue accessed on 20 May 2022
8
1, January–February 2019, pp. 15-25, Rohan George (24 November 2015) < cis-india.org >
<https://doi.org/10.1016/j.bushor.2018.08.004> accessed accessed 24 May 2022
9
on 20 May 2022 Ibid

8 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
like Madhya Pradesh, Uttar Pradesh, Punjab, 2. ARTIFICIAL INTELLIGENCE
and Rajasthan have also adopted such use of IN PREDICTIVE AND
technology, especially for facial recognition.
PREVENTIVE POLICING
1.4 Challenges to Predictive Policing According to Lau, predictive policing “is the
Every pro has a con, similarly implementing use of algorithms to analyze massive amounts
the latest technology will have its implications, of information to predict and help prevent
where public and private entities are suggested potential future crimes”.13
to make AI more inclusive addressing issues of
bias and making it more open and accessible to 2.1 Various applications of AI in Policing
a large number of masses. Here, enforcement • According to the report published by
agencies are aiming to remove human Carnegie endowment14, 56 countries are using
interventions to reduce the number of errors the technology of face recognition where EU is
that will always be a part of it making laws and one of them where 40 civil organizations have
supporting the enforcement system. demanded to ban predictive policing and the
European Commission has proposed to
Rather than looking for the criminals in the regulate such an act.
historical data, predictive analysis is done to • 15 states in India are the most
find out people who were more police, there is surveillance cities named, including Delhi,
a legitimized discrimination behind this Indore, Chennai, and Hyderabad with the
mathematical analysis, as lower caste and greatest number of CCTV cameras. In IHS
religious minorities have faced more violence Markit’s latest report Delhi topped the list
and discrimination despite constitutional followed by Chennai and Mumbai. In 2021,
remedies.10 Kolkata police announced the installation of
2500 CCTV which will be to spot bikers with
In the case of “Ankush Maruti Shinde v. State helmets and vehicles in illegal parking. As per
of Maharashtra”11, due to their ethnicity, the the increasing facial recognition, it seems by
police assumed they were guilty and sentenced 2024 this number will multiply by 6 times and
them to 16 years in solitary prison. Other will get close to China’s surveillance system.
groups, including as Muslims, Dalits, and Further Telangana is planning to build an
Adivasis, also face prejudice in the criminal integrated police center in banjara hills costing
justice system. Predictive analysis also uses a 800 crores for law enforcement to view
huge amount of personal data violating the surveillance footage across the city in real-time.
right to privacy as envisioned under the case of • Other countries are also similarly
“Justice K.S. Puttaswamy (Rtd.) v. Union of adopting technology at a fast pace like China,
India”12, where the use of such data must be the US, the UK, Japan, Brazil, and Singapore,
sanctioned by law, and under a democratic In US cities like Detroit, Orlando, and
society, procedural safeguards against such Washington DC are testing the use of facial
misuse must exist, and appropriate interference recognition videos for Safety and security
is a necessary issue. Such use of AI in law and according to the report by Georgetown
order needs greater transparency, increased University Centre on Privacy and
supervision, and adequate mechanism. Any Technology.15 China surpasses the US market
government data collection that necessitates in city surveillance networks, with one camera
proactively informing a person about how their for 4.1 people, whereas India has the world’s
data is used should be subjected to a rigorous largest biometric database which can easily
evaluation as this data is not only gathered by scan irises, fingerprints, and photos of citizens.
public authorities but also by commercial This can be a challenge to the privacy of people
organisations.
13
T. Lau, ‘Predictive Policing Explained’, Brennan Center
for Justice, 2020 <https://www.brennancenter.org/our-
work/research-reports/predictive-policing-
explained> accessed 25 May 2022
14
Steven Feldstein, ‘The Global Expansion of AI
10
Ramchandran Murugesan, ‘Predictive Policing in India: Surveillance’ <
Deterring crime or Discriminating Minorities?’ https://carnegieendowment.org/2019/09/17/global-
<https://blogs.lse.ac.uk/humanrights/2021/04/16/predictiv expansion-of-ai-surveillance-pub-79847> accessed on 24
e-policing-in-india-deterring-crime-or-discriminating- May 2022
15
minorities/> accessed on 24 May 2022 Clare Garvie & Laura M. Moy, ‘Face Surveillance In The
11
(2009) 6 SCC 667 United States’ America Under Watch
12
(2017) 10 SCC 1 <www.americaunderwatch.com> accessed on 24 May 2022

9 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
as it can be misused by our authorities which • Thirdly, the UN will discuss such
they claim to be secure and not meant for complete concerns and identify any minor or
monitoring people in general. Singapore has significant threats.
also installed 80,000 smart cameras which are
capable of analytics through running videos 2.2.2 US in AI Policing
which a quite progressive. The US is amongst the top AI users, where AI
applications are discussed for competition,
2.2 Global Best practices employment, data protection, privacy, legal
2.2.1 UN on AI in Law Enforcement16 enforcements, sectoral growth, etc, here also
A new pattern of predicting crimes and facial recognition cameras and predictive
criminal behaviour is now being aimed through policing play a major role but keeping in mind
AI technologies which can change the face of the viewpoint of the public, regulatory reforms
law enforcement in our legal system. This are being made to regulate the AI enforcement
technique is being used through a large amount system in maintaining data privacy. There are
of data and robotics which uses data analysis to no general guidelines for applying AI, but the
forecast high levels of crimes. This is also White House Office of Science and
resulting in a threat to the international Technology has listed 10 principles to take into
community, violating their human rights and account when drafting regulatory and
privacy. State authorities may be using racial or nonregulatory procedure. A few of these
ethnic profiling to target particular groups or principles are: use transparent risk assessment
expand the capability of monitoring through and risk management approaches; consider all
the use of technologies like face or fingerprint societal costs, benefits, and other externalities
recognition. when deciding whether to develop and deploy
AI; consider the ethical implications of using
The majority of nations are in the emerging AI; considerations when developing to adapt to
stage and they are eager to learn from one the evolving nature of AI, pursue performance-
another, particularly from the commercial based and adaptable techniques, determine the
sectors that are putting AI and robots into proper disclosure and openness levels to win
practise. As of 2017, UNICRI developed a over the public. The US generally remains
centre concentrating on criminal justice, crime highly on investments in AI and related
prevention, and security in general. This technologies, and it can be overtaken by China
indicates that AI technologies are still in their as the US does have ample resources but it
infancy. seems not to be utilized efficiently.

This was the first gathering in this area for law 2.2.3 China in AI Policing
enforcement, and the UN and INTERPOL China is already a cutting-edge global leader in
jointly convened the event to examine surveillance through AI, many Chinese firms
innovations, possibilities, and concerns related have built software that uses AI to sort and
to AI and robots and to handle new issues. organize data on residents by upgrading their
Both the use of AI to catch criminals and the tools. China has blanketed the cities with
use of AI by criminals were covered in the camera surveillance and planning to do the
discussion. The role of the UN under such same across the rural areas. Recently the city
surveillance gadgetry would be threefold: named Handan launched 3 futuristic robots
• First and foremost, more which are equipped with AI and facial
conversations and debates will be held to recognition technologies, they help in
increase public understanding of AI, patrolling, controlling traffic accidents, and in
particularly in poorer nations where AI has a providing relevant information, equipped with
negative social and economic impact. a navigation system. They all are facilitated
• Second, they may optimise the with big data technology which was the part of
advantages of AI-driven technology while 13th Beijing 5 years plan.17 In 2017 Zhengzhou
taking into account the SDGs alongside other police were equipped with smart eyeglasses for
UN programmes. screening down people who are involved in
criminal activities, which helped them to find

17
Bryan Ke, ‘China Deploys First AI Robot Police That
16
Iraki Beridze & Summer Walker, ‘Artificial Intelligence Tackles Criminals and Now We’re Living in ‘Black
in Policing’ <https://globalinitiative.net/analysis/artificial- Mirror’ <https://nextshark.com/china-ai-robot-police/>
intelligence-in-policing> accessed on 24 May 2022 accessed on 26 May 2022

10 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
out through 10,000 identity databases at a lower and higher judiciary. Surprisingly IBM
speed of 100 milliseconds. Similarly, 20 robot ‘ROSS’ which is having data mining and
million CCTVs were being installed all over identifying techniques will be used for back-
China just to identify criminals through end office work in litigation and arbitration
tracking facial recognition from their database. process such as research work, and data
China is consistently growing by fuelling its storage, this will not only help our legal system
desire to become the world leader in AI in research but also in handling significant
technology.18 pendency of cases plus on counseling clients
and taking on complex issues, in keeping track
Although the US remains ahead amongst other of the work being done by creating invoices
countries in AI forces, if China implements its which enables transparency to firms, clients,
strategies and policies effectively, it will soon lawyers, and auditing authorities.
replace the US. Through this, we can imagine
the thriving of AI, data science, and machine 2.3.1 AI enforcement Agencies
learning. To find criminals and solve such cases, Indian
law enforcement agencies are turning to
2.2.3 UK in AI Policing artificial intelligence (AI) technologies.
Facial recognition and predictive policing Gurugram-based Staqu Technologies, which
through AI are the emerging trends in the UK was founded in 2015, has developed a
since 2014, this helps in capturing images of technology stack that includes “sophisticated
criminals by comparing them against the stored image analysis, facial recognition, and text
database, mostly this has been used by processing tools for entity recognition and
metropolitan police and south whale forces. summarization APIs”. Recently this company
2019 report has found that 70% of the public has launched an AI-enabled application
supports the investigation by this technology ‘Artificial Intelligence-based Human Efface
and 55% want it to be done in special Detection’ with the help of Rajasthan police to
circumstances only. Through the shared Police identify criminal identity and information of
National Database, which houses a vast missing persons, similarly, the State of Punjab
amount of information including pictures and used this technology. Crime and criminal
prior criminal records, all territorial police in tracking systems can easily help us to search
the United Kingdom do have access to their applications through biometric information
history.19 like fingerprints, and face recognition, by this
they have digitized more than 50,000 criminal
The worry here is AI allows cost savings in records and have solved 100 cases. It can give
different spheres, which will decrease other 95% accuracy in facial recognition as to gender,
measurable benefits like community and facial changes, and facial occlusion. Smart
relationship building, this may also increase glasses were also created by Staqu
efficiency gains misleading to produce Technologies. These glasses have a built-in
unintended results. Second, the utilisation of camera for face recognition to identify persons
historical data will make racial and gender in a crowd, as well as data feeds that show
prejudice worse owing to a lack of images of objects onto the user's screen. Apart
transparency, which may jeopardise the idea of from this Kerala government use drones for
policing by consent. facial recognition and identification using AI
software, generally used for high-security
2.3 AI implementation in India's zones, near airports, and at traffic, junctions to
administration of the criminal justice system identify criminal suspects.20
According to the National Judicial Data Grid
(NJDG), over crores of cases are pending in 2.3.2 Government Initiatives
AI holds the potential to protect our citizens
from any kind of criminal acts and other
18
Holly Chik, ‘Surveillance State: 18 of the World’s 20 related threats through better surveillance,
Most Monitored Cities are in China’ faster grievance redressal, and smart policing,
<https://www.scmp.com/news/china/society/article/30946
66/surveillance-state-18-worlds-20-most-monitored-cities-
20
are-china> accessed on 26 May 2022 Abhishek Baxi, ‘Law Enforcement Agencies in India
19
Will Grimond & Asheem Singh, ‘A Force for Good? Are Using Artificial Intelligence to Nab Criminals’
Results from FOI requests on artificial intelligence in the <https://www.forbes.com/sites/baxiabhishek/2018/09/28/l
police force’ aw-enforcement-agencies-in-india-are-using-artificial-
<https://www.thersa.org/globalassets/reports/2020/a-force- intelligence-to-nab-criminals-heres-how/> accessed on 26
for-good-police-ai.pdf > accessed on 26 May 2022 May 2022

11 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
which have been taken by the government 3. ETHICAL AND LEGAL ISSUES
agencies through cutting edge technology to OF AI POLICING IN INDIA
improve citizen safety. With the use of artificial intelligence, crime
Following are a few operative AI initiatives21- may be predicted, prevented, and recognised
• Government of UP- recently UP and while it has already happened or is happening.
Staqu jointly employed 700 video analysis Around the world, law enforcement agencies
cameras across 70 prisons to have a 24/7 have been utilising AI-powered tools to
surveillance and data feed of the visitors to the investigate crimes and, on occasion, even
prison, crowd analysis, violence and fights in attempt to forecast them. To locate the holy
prison, facial recognition and analysis, grail of policing—preventing crime before it
behavioural analysis and unauthorized occurs—the police are using predictive policing
intrusion. The real-time insights can be easily tactics employing data analytics, powerful
visualized by the police officials through a computers, and intuition.23
mobile application.
• Government of Telangana- In 2017 W. Perry classified Predictive policing
the Telangana police launched a smart methods into four categories including “those
Robocop with inbuilt cameras, GPS sensors, which predict crimes, those which predict
temperature sensors, and ultrasonic readers. offenders, those which predict the
This was designed by the Hyderabad company perpetrator’s identities and those which predict
H-Bot to assist police officials in maintaining victims”.24 Some concerns regarding the nature
law and order, it has emergency flashlights and of prediction in a time where data collecting
charging points with the technology of thermal and analysis are prevalent have been raised by
imaging which can easily extend help for each of these approaches. Ferguson observes
security purposes, in interacting with people, that “foretelling the accurate identity of the
taking complaints, and solving queries. future human ‘criminal’ presents a far more
• Government of Haryana- In 2018 troubling prediction”. The studies and civil
Haryana government had an agreement with liberties concerns may both be based on
city cops to develop an application in historical data with statistically significant
replacement of the existing application correlations, but they are not the same.25
Harpeth for road-related grievances to help in Most of the predictive softwares used by police
receiving complaints and analyzing same which and other agencies are basically hollow
can also reduce manpower and cost involved in promises. This is demonstrated by the
the process. programme ‘PredPol,’ which was created by a
• Government of Maharashtra- In significant software business and is regarded as
2015, they invested Rs 800 crore plan to the industry leader in predictive policing in the
secure cybercrime to prevent, predict, and US. Predpol identifies the locations and times
detect cybercrimes in real-time. This works on when specific crimes are most likely to occur so
big data and algorithms for collecting data on that these areas may be patrolled to prevent
crime using available online sources.22 these crimes from occurring.26 There are

In Asia Pacific India stands 3rd after Singapore


and Hong Kong and 19th on a global level, it is 23
Justin Jouvenal, ‘Police are Using Software to Predict
one of the most prepared economies with the Crime. Is it a ‘Holy Grail’ or Biased Against Minorities?’,
WASH. POST (Nov. 17, 2016),
government support in AI technologies but <https://www.washingtonpost.com/local/public-
lacks creativity and innovations from the safety/police-are-using-software-to-predict-crime-is-it-a-
developed countries, where it is trying to build holy-grail-or-biased-against-minorities/2016/ 11/
up such a tech-savvy environment through 17/525a6649-0472-440a-aae1-
b283aa8e5de8_story.html?utm_term=.72a9d2eb22ae>
emerging start-ups with skilled professionals in accessed 23 May 2022
AI, language processing and machine learning. 24
W. Perry, et al, 2013, ‘Predictive Policing: The Role of
Crime Forecasting in Law Enforcement Operations’,
RAND Corporation, p.14.
<https://www.rand.org/content/dam/rand/pubs/research_re
21
INDIAai, ‘Five AI Initiatives for Public Safety in India’ ports/RR200/RR233/RAND_RR233.pdf> accessed on 25
< https://indiaai.gov.in/article/five-ai-initiatives-for-public- May2022.
25
safety-in-india> accessed on 26 May 2022 Andrew G. Ferguson, ‘Policing Predictive Policing’, 94
22
‘Maharashtra government to come up with `predictive WASH. U. L. REV. 1109 (2017).
policing policy’, PTI, <https://openscholarship.wustl.edu/law_lawreview/vol94/i
<https://indianexpress.com/article/india/maharashtra- ss5/5> accessed 23 May 2022
26
government-to-come-up-with-predictive-policing-policy- “Overview”, PredPol <https://www.predpol.com/about/>
5097896/> accessed on 26 May 2022 accessed 23 May 2022

12 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
several difficulties that illustrate the necessity and class biases according to judicial ethics. For
to carefully evaluate any excessive usage of AI instance, the parties would hope that judges
and that there are times when those advantages were able to properly evaluate the entire case,
appear to exceed those dangers. In order to balancing pros and drawbacks before coming to
highlight and solve the enormous ethical, legal, a judgement when granting bail. However,
and social concerns that may develop in the research by psychologists has revealed that
future as we move towards an automated judges only mentally run an organised checklist
future with limitless potential for AI in law of red flags. If the defendant's account raises
enforcement, Ang and Feinholz argue.27 any of those red flags—previous convictions,
links to the community, or the prosecution's
3.1 Ethical Issues request—the court will halt and refuse bail.32
Hannah Fry28 has observed that “for all the AI also applies similar structured algorithms to
positive impacts that AI may have on the give a required decision.
criminal justice system, there will invariably be • In case of facial recognition
endless examples of unfairness engendered by technologies there is a major concern regarding
algorithms”.29 Algorithms created to simulate the “possibility for false identification.”
intelligence will always make blunders. When Consequences of false identifications can be
law enforcement uses them, they could be severe for those who are wrongfully identified
pointless or unjust. There are concerns on how as a suspect/accused. According to Hanna Fry,
and when these tools have to used. Some of “the more faces the algorithm searches
these issues are highlighted below: through, the greater the chance of it finding
• Confirmation Bias: The police are two faces that look similar”.33
likely to predict and detect more crimes at a • Objectives: When police or any agency
place with the use of AI, than they would have decides to implement and utilize AI as a
done otherwise30 regulatory mechanism for crime than the
• Most of the predictive technology and priorities must be set out clearly. Is it keeping
forecasting softwares and the algorithms which crime as low as it can be? or prioritising the
are used in them are proprietary. For a person innocent's right to freedom? What portion of
like a judge or police officer the technological one would you give up for the benefit of the
knowhow of the algorithms is not available. other?34
The accused person may not receive a fair trial
and be denied their constitutional rights if 3.2 Social Challenges
there is no way to analyse these algorithms and Some of the social concerns associated with use
how they produce their conclusions.31 of AI in policing includes:
• The great majority of judges strive to • Discrimination. Accuracy of Artificial
be as impartial as possible in the courtrooms, Intelligence is dependent on the learning from
yet evidence has consistently demonstrated fed data. However, when the data fed to AI is
that they engage in discrimination. A judge's biased then it will results in bias in its
judgement must be free from racial, gender, decisions making and results. Bias may occur
based on the factors like gender, race, colour,
27
Tee Wee Ang and Dafna Feinholz, ‘Working for, not language, etc.35 Furthermore, if just data from
against, humanity’, The UNESCO Courier, July- one ethnic group was utilised for the learning
September 2018, p. 29 stage, facial recognition software may not be
<https://unesdoc.unesco.org/ark:/48223/pf0000265211>
accessed 25 May 2022 able to recognise members of that group.36
28
She is a British mathematician and an expert on • In many situations, AI has the
computer science and human behavior. She has identified potential to be a very potent tool. AI may
numerous ethical issues raised by the use of AI for crime expose characteristics about people that they
analysis in her book.
29
Hannah Fry, Hello World: Being Human in the Age of
the Machine (New York, NY: W.W. Norton, 2018) at
32
330-332 Supra note 29
33
<http://home.ustc.edu.cn/~ustcsh/py2016/data/Warren%2 Ibid
34
0Sande,%20Carter%20Sande- Ibid
35
Hello%20World!_%20Computer%20Programming%20for% Tolga Bolukbasi et al, ‘Man is to Computer Programmer
20Kids%20and%20Other%20Beginners- as Woman is to Homemaker? Debiasing Word
Manning%20Publications.pdf> accessed 25 May 2022 Embeddings’ (2016)
30
Ibid <https://arxiv.org/pdf/1607.06520.pdf> accessed 28 May
31
Danielle Keats Citron, ‘Technological Due Process’, 2022
36
WASH. U. L. REV. 1249 (2008) ‘Is facial recognition technology racist?’, The Week UK
<https://openscholarship.wustl.edu/cgi/viewcontent.cgi?arti (27 July 2018), <https://www.theweek.co.uk/95383/is-
cle=1166&context=law_lawreview> accessed 23 May 2022 facial-recognition-racist> accessed 28 May 2022

13 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
would prefer to keep private or aren't even • Select the pretrial release option.
aware of themselves by inferring those
characteristics based on other data. If not Rowena Rodrigues41 states that the
handled appropriately, it may also be used to employment of AI in police raises a number of
push a large number of people in a particular legal difficulties. Some of these are related to:
way, undermining democratic values. AI runs • The structure and makeup of AI itself
the danger of preserving biases in the data it • Problems with the implementation
uses to learn from. This might maintain • Using AI
systemic inequality and strengthen the status
quo. Given how challenging it may be to These issues are interconnected and are
comprehend AI models, this distinction may common of all computer technologies such as
be difficult to notice. privacy/data protection, transparency, fairness,
• AI applications can’t set aside the accountability, etc.
human component entirely. On the basis of
data the algorithms usually derive results or a 3.3.1 Data Generation, Vulnerabilities and
number, but the interpretation of such results Responses
or number is for the user. Završnik highlights The development of AI may be abused to
this with an illustration that “at what level of provide data that seems incredibly real. Such
probability of recidivism should a prisoner be false information may be employed in social
granted parole? Whether this threshold ought engineering techniques. The users have learned
to be a 40 percent or an 80 percent risk of to manipulate the images using softwares like
recidivism is an inherently ‘political’ decision Adobe Photoshop. With the help of AI they
based on the social, cultural and economic can even manipulate media such as sound and
conditions of the given society”.37 video at a large scale. The FakeApp tool
• Increasing Crime using AI: The released by Reddit allowed users to do
criminals have inclined to use AI for engaging “generate fake videos using deep learning
Social Engineering, Phishing, Vishing, for networks that rely on a technology known as
eliciting information”.38 Matt Chessen throws autoencoders”.42 Such data generation and
light on “use of AI target messages so they vulnerability in data is likely to cause serious
become more convincing to certain people concerns for AI policing when the data is
based on their socio-demographic related to a crime suspect or a future criminal.
characteristics or psychological traits”.39
• The accuracy of predictive police 3.3.2 Privacy Issues
algorithms depends on the quality of the data The right to decide when and to whom to
they use, and skewed data institutionalises release personal information can be referred to
prejudice against minorities. It's also possible as privacy. This privilege is significantly
that the government monitors its residents threatened by contemporary artificial
using predictive policing technologies.40 intelligence capabilities and the extensive
collecting of private data. Predictive policing
3.3 Legal Concerns algorithms employ enormous amounts of
Utilizing AI for proactive policing, the law personal data, which undoubtedly has a
enforcement will supposedly be able to: negative impact on the right to privacy. The
• Predict the criminal areas. privacy rights outlined in Justice K.S.
• Predict who will commit the crime. Puttaswamy (Retd.) v. Union of India are
violated by the use of personal data that is
opaque.43
37
A. Završnik, ‘Algorithmic justice: Algorithms and big
data in criminal justice settings’, European Journal of
Criminology, (2021) 18 (5), pp. 623–642. 3.3.3 Opacity with respect to Algorithms
DOI: 10.1177/1477370819876762
38
“Vishing”, Security Through Education,
<https://www.social-engineer.org/framework/attack-
41
vectors/vishing/> accessed 28 May 2022 Rowena Rodrigues, Legal and human rights issues of AI:
39
Matt Chessen, ‘The Madcom Future: How Artificial Gaps, challenges and vulnerabilities, Journal of
Intelligence Will Enhance Computational Propaganda, Responsible Technology, Volume 4, 2020,
Reprogram Human Culture, and Threaten <https://doi.org/10.1016/j.jrt.2020.100005> accessed on
Democracy...and What Can Be Done About It’, The 28 May 2022
42
Atlantic Council (1 September 2017), Gaurav Oberoi, ‘Exploring DeepFakes’, Hacker Noon (5
<https://www.scribd.com/document/359972969/The- March 2018), < https://hackernoon.com/exploring-
MADCOM-Future> accessed 27 May 2022 deepfakes-20c9947c22d9> accessed on 27 May 2022
40 43
Supra Note 10 (2017) 10 SCC 1

14 | Page https://jlai.nfsu.ac.in/
JLAI
Volume: 2, Issue: 1
June 2023
E-ISSN: 2584-1661
Furthermore, it is impossible to determine the • All public procurement agreements
precise workings of predictive policing should include a provision requiring the
algorithms since law enforcement agencies are retroactive deconstruction of the algorithm in
excluded from disclosure under the Right to order to evaluate the factors that impacted the
Information Act of 2005. Due to this opacity, model's projections.
many people, including myself, believe that • To guarantee conformity to this
predictive policing is governmental monitoring regulatory framework, a defined system of
that is disguising itself as internal security.44 inspection and oversight (such as a
commission, taskforce, committee, board, etc.)
3.3.4 Gap in Legal Framework is required. These ethical committees ought to
There isn't a comprehensive data protection be interdisciplinary and comprise a mix of
legislation in force in India right now. Also, the professionals, subject-matter specialists,
regulatory guidelines on use of AI by law academics, and maybe ordinary or lay people.
enforcement is still to see the light of the day. • To guarantee that all specialists and
stakeholders are represented, the strategy
4. MEASURES FOR ACCURATE should be collaborative and bridge disciplinary
AND EFFECTIVE AI POLICING boundaries. This body ought to advise specific
IN INDIA law enforcement organisations on best
practises, strategies, and overall policies
For governments whose law enforcement
pertaining to the employment of algorithms.
agencies want to deploy AI, there are several
options accessible. These options call for
proactive legislation infused with a dedication 5. CONCLUSION
to the minimal criteria for openness, oversight Body cameras that fit properly and the usage of
mechanisms, and multidisciplinary AI to create crime reports and forecasts are
collaboration. The government supporting law examples of how artificial intelligence is
enforcement agencies who are experimenting becoming more prevalent in police. However, a
with AI also has a growing body of research at much more contentious and far-reaching
its disposal on the best practises for algorithmic application of this technology is using it to
decision-making. ascertain if a crime was actually committed.
The results of the study conducted by For instance, artificial intelligence is utilised in
Alexander Babuta, Marion Oswald, and the developing subject of predictive policing to
Christine Rinik for the Royal United Services anticipate future crime hotspots, such as where
Institute in the UK offer critical insight into and when a specific kind of crime is most likely
the policies that could be implemented to to occur. The performance of AI is based on an
increase the accountability, precision, and algorithm designed by a human being and it
effectiveness of AI policing, regardless of performs on the basis of data fed by the
jurisdiction. Some of these measures could be humans. Human are bound to make mistake,
adopted in India as well. These include45: so will the AI. However, it is not ethical to give
• It is vital to create guidelines and rules that authority to a computer when mistakes
of conduct that specify how law enforcement made by AI result in someone losing their
should test and employ algorithmic freedom or being charged with a crime. To
technologies. address the ethical, social, and legal issues
• The use of algorithmic tools by police raised by the use of artificial intelligence in
forces must adhere to minimal criteria set by a police, the government must develop
regulatory framework, particularly in relation technological, legal, and algorithmic solutions.
to applicable data protection laws, the AI
system's openness and understandability, and
adherence to administrative law and human
rights principles.

44
Supra Note 10
45
A. Babuta, M. Oswald, & C. Rinik, ‘Machine Learning
Algorithms and Police Decision-Making: Legal, Ethical and
Regulatory Challenges’. Whitehall Reports, no. 18, vol. 3,
2018, Royal United Services Institute, London.
<https://rusi.org/publication/whitehall-reports/machine-
learning-algorithms-and-police-decision-making-legal-
ethical> accessed on 24 May 2022

15 | Page https://jlai.nfsu.ac.in/

You might also like