What is technology disruption?
A new technology that changes
The way things are done
Expectations and behaviors
Market or industry
Uber/Grab vs taxi business
Airbnb vs hotel business
Netflix vs TV, DVD, cinemas
Blockchain vs financial institutions
Google vs books, libraries
ChatGPT vs Google
How can technology disrupt the Law?
DoNotPay
https://donotpay.com/
Bizibot
https://www.btlnet.co.uk/digital-transformation-solutions-overview/bizibot
WeVorce
https://www.wevorce.com
OCBC Online Will Generator
https://www.ocbc.com/personal-banking/lifegoals/willgenerator/#/
Transcripts
https://fortherecord.com/transcript-express/
Law firms using Blockchain
https://www.raconteur.net/technology/blockchain/blockchain-revolutionising-legal-sector/
LawGeex AI vs human lawyers, on reviewing standard contracts
Accuracy
Time taken to review contract
Courts used technology to deliver justice during COVID
But litigants with resources had the advantage
https://www.pewtrusts.org/en/research-and-analysis/reports/2021/12/how-courts-embraced-technology-met-the-pandemic-challenge-and-revolutionized-their-operations
Disruption has no ethics. But we do.
Must bridge the Digital Divide
Parties may need help
https://www.ncsc.org/__data/assets/pdf_file/0026/53738/PPP-Technology-Digital-Divide-Considerations.pdf
Using AI for judgments
Benefits
Exclude legally irrelevant factors
Standardise decisions
Help courts to prioritise cases, increase efficiency
Encourage parties to settle if success is predicted to be unlikely
https://www.cppr.in/articles/predictive-justice-using-ai-for-justice-introduction
Risks
Over reliance on machine decisions
Algorithmic bias: bad history = bad future
Bad data feedback into bad data
Inability to explain decisions: bad for rule of law
https://www.suls.org.au/citations-blog/2020/9/25/predictive-judicial-analytics-implications-for-rule-of-law-and-the-legal-profession
Cases of AI bias in justice system
COMPAS (for probation)
Black defendants were 2x wrongly predicted to be high risk of recidivism
White defendants were 2x wrongly predicted to be low risk of recidivism
Black defendants were 2x more likely to be misclassified as high risk of violent recidivism
White violent recidivists were 63% more likely to have been misclassified as low risk
https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm
Public Safety Assessment (for bail)
More likely to recommend high bail for males than females
https://venturebeat.com/ai/study-finds-crime-predicting-judicial-tool-exhibits-gender-bias/
How does this type of AI work?
How does this type of AI work?
Where did the ML (Machine Learning) come from?
Garbage In = Garbage Out
Always check what’s going in and what’s coming out
Rand Report on Protecting Individuals using tech in the Justice
Technology Disruption Digital Inclusion and the Law.pdf
1. Technology Disruption,
Digital Inclusion,
and the Law
Benjamin Ang – Co-Chief Data Officer, AI Singapore
Deputy Head, Centre of Excellence for National Security
Twitter.com/benjaminang
2. What is technology disruption?
• A new technology that changes
• The way things are done
• Expectations and behaviors
• Market or industry
• Uber/Grab vs taxi business
• Airbnb vs hotel business
• Netflix vs TV, DVD, cinemas
• Blockchain vs financial institutions
• Google vs books, libraries
• ChatGPT vs Google
10. LawGeex AI vs human lawyers, on
reviewing standard contracts
Accuracy Time taken to review contract
0
200
400
600
800
1000
1200
1400
Human AI
Seconds
Seconds
78
80
82
84
86
88
90
92
94
96
Human AI
% Accuracy
% Accuracy
12. But litigants
with
resources had
the advantage
https://www.pewtrusts.org/en/res
earch-and-
analysis/reports/2021/12/how-
courts-embraced-technology-met-
the-pandemic-challenge-and-
revolutionized-their-operations
14. Must bridge
the Digital
Divide
Parties may need help
https://www.ncsc.org/__data/asse
ts/pdf_file/0026/53738/PPP-
Technology-Digital-Divide-
Considerations.pdf
15. Using AI for judgments
Benefits
• Exclude legally irrelevant factors
• Standardise decisions
• Help courts to prioritise cases, increase
efficiency
• Encourage parties to settle if success is
predicted to be unlikely
• https://www.cppr.in/articles/predictive-
justice-using-ai-for-justice-introduction
Risks
• Over reliance on machine decisions
• Algorithmic bias: bad history = bad future
• Bad data feedback into bad data
• Inability to explain decisions: bad for rule
of law
• https://www.suls.org.au/citations-
blog/2020/9/25/predictive-judicial-analytics-
implications-for-rule-of-law-and-the-legal-
profession
16. Cases of AI bias in justice system
COMPAS (for probation)
• Black defendants were 2x wrongly
predicted to be high risk of recidivism
• White defendants were 2x wrongly
predicted to be low risk of recidivism
• Black defendants were 2x more likely to be
misclassified as high risk of violent
recidivism
• White violent recidivists were 63% more
likely to have been misclassified as low risk
• https://www.propublica.org/article/how-
we-analyzed-the-compas-recidivism-
algorithm
Public Safety Assessment (for bail)
• More likely to recommend high bail for
males than females
• https://venturebeat.com/ai/study-finds-
crime-predicting-judicial-tool-exhibits-
gender-bias/
22. Rand Report on Protecting Individuals
using tech in the Justice System
• Maintain quality and integrity of data used by the criminal justice system
• Train justice system staff on how to use the technology correctly
• Study how the tools (e.g., risk assessment tools) are performing
• Study how to protect the data of people in the system (victims, plaintiffs,
witnesses)
• https://www.rand.org/pubs/research_reports/RR1748.html
• Also: provide all participants with the tools and help needed to access justice
23. What will you do if you
see a new disruptive
technology?
24. Technology Disruption,
Digital Inclusion,
and the Law
Benjamin Ang – Co-Chief Data Officer, AI Singapore
Deputy Head, Centre of Excellence for National Security
Twitter.com/benjaminang