MindMap Gallery Artificial Intelligence Machine Discrimination and Countermeasures
Research on the fairness of machine learning, focusing on the issue of discrimination. The content includes: artificial intelligence is ushering in a development wave, artificial intelligence decision-making is becoming increasingly popular, machine bias/prejudice (Machine Bias) cannot be ignored, and three major issues in artificial intelligence decision-making, Construct technical fairness rules and achieve fairness by design.
Edited at 2023-08-01 16:37:05This is a mind map about bacteria, and its main contents include: overview, morphology, types, structure, reproduction, distribution, application, and expansion. The summary is comprehensive and meticulous, suitable as review materials.
This is a mind map about plant asexual reproduction, and its main contents include: concept, spore reproduction, vegetative reproduction, tissue culture, and buds. The summary is comprehensive and meticulous, suitable as review materials.
This is a mind map about the reproductive development of animals, and its main contents include: insects, frogs, birds, sexual reproduction, and asexual reproduction. The summary is comprehensive and meticulous, suitable as review materials.
This is a mind map about bacteria, and its main contents include: overview, morphology, types, structure, reproduction, distribution, application, and expansion. The summary is comprehensive and meticulous, suitable as review materials.
This is a mind map about plant asexual reproduction, and its main contents include: concept, spore reproduction, vegetative reproduction, tissue culture, and buds. The summary is comprehensive and meticulous, suitable as review materials.
This is a mind map about the reproductive development of animals, and its main contents include: insects, frogs, birds, sexual reproduction, and asexual reproduction. The summary is comprehensive and meticulous, suitable as review materials.
Artificial Intelligence: Machine Discrimination and Countermeasures
introduction
Interfere with and influence people's real lives
algorithm
Big Data
AI
algorithm
what we do in the online world
Various trails and activities
Internet habits
Shopping history
GPS location data
Transformed into ratings and various predictions for us
us
Presence in the network
increasingly dominated by algorithms
We increasingly live in a score-based society
credit score
crime risk score
Ignoring the issues of algorithmic and machine discrimination
Automated decision-making system
application
Education, employment, credit, loans
Insurance, advertising, medical care, public security
criminal justice process
lead to
The problem of algorithmic and machine discrimination becomes increasingly prominent
Algorithms and mathematics are not necessarily objective
Programmer’s subjective judgment and choice
codification of rules
opaque
Inaccurate
unfair
Unable to censor
Can it be outsourced to technology?
Morality
ethics
law
How to achieve and ensure the fairness of automated decision-making
Artificial intelligence is ushering in a wave of development
Since 2010
Artificial intelligence ushered in a new wave of development
Related entrepreneurship, investment and mergers and acquisitions have significantly strengthened
October 2016
National Artificial Intelligence Strategy
U.S. National Artificial Intelligence R&D Strategic Plan
British "Robotics and Artificial Intelligence"
Artificial Intelligence Decision-Making is Growing in Popularity
algorithm
Decide
What news do people see?
What ads do you receive?
what song did you hear
about
People’s online presence (digital presence)
artificial intelligence decision-making
Shopping recommendations
Precision advertising
credit assessment
crime risk assessment
Issues that cannot be ignored
Delegating decision-making to artificial intelligence
Can machines be impartial?
How to ensure fairness is achieved?
Machine Bias cannot be ignored
Google Photos
Black people were mistakenly labeled as "gorillas"
Flickr automatic tagging system
Photos of black people mistakenly labeled as "apes"
Microsoft AI chatbot Tay
Being "taught bad"
Google algorithm discrimination
Black names are more likely to appear in ads suggesting a criminal history
Men see more high-paying job ads than women
Amazon
Concealed shipping costs for its own and partners’ products
Crime risk assessment algorithm COMPAS
White people are more likely to be assessed as having a low crime risk
Black people assessed as high crime risks
Very unreliable in predicting future crime
Discrimination against ethnic minorities
The probability of identifying a surname as false is high
Three major issues in artificial intelligence decision-making: fairness, transparency and accountability
Is the algorithm fair by default?
misunderstanding
Algorithmic decisions tend to be fair
reason
Mathematics is about equations
rather than skin color
math cleaning
worship of data
human social things
Use math
quantification of services
objectification
Need answers to questions
Can fairness be quantified and formalized?
Can it be translated into operational algorithms?
Are there risks associated with fairness being quantified as a matter of calculation?
Who determines the considerations for fairness in AI?
How to make AI have a concept of fairness and be aware of discrimination independently?
The Transparency Dilemma of Algorithms as "Blackbox"
Algorithmic opacity
Three forms of opacity
Opacity arising from commercial or state secrets
Opacity due to technical illiteracy
Opacity arising from algorithm characteristics and measurements
How to hold the algorithm accountable?
Need to solve the problem
What do people need to review?
How to judge whether an algorithm complies with existing legal policies?
How can algorithms be scrutinized in the absence of transparency?
Construct technical fairness rules and achieve fairness through design (Fairness by Design)
government level
White House report
Included in the National Artificial Intelligence Strategy
It is recommended that practitioners receive ethics training
british science and technology council
Call for the establishment of a specialized artificial intelligence committee
industry level
Proposed the concept of "Equality of Opportunity"
Matthew Joseph
Introducing the concept of "Discrimination Index"