Please enable JavaScript.
Coggle requires JavaScript to display documents.
AI and ML - Coggle Diagram
AI and ML
Cybersecurity Risks
Examples of Risks
Adversarial attacks, data poisoning, model inversion and extraction, privacy concerns, unintended consequences and bias, lack of interpretability and explainability, malicious use of AI
A business is installing an AI system by the top sowftawre company. They are pressed for time and have the option to trust the system or test it and get a second opinion before launching the system
In this scenario the ethical decision would be to first test the software and get a second opinion. This will ensure it is free from bias and works properly. Getting a second opinion such as utilizing ethics by committe will help to ensure it is ready for use.
The unethical decision would be to trust the system is free from biias and error without testing it before use.
Looking at the business scenario vice versa, upon finishing the sowftware company could realize they made a slight error in the programming. They have the option to take the time to correct it, or trust that it will not affect the system that much
In this scenario the ethical decision would be to go back and correct the error, even if it was just a slight one.
Besides just the chance of legal consequences, utilizing virtue ethics, being honest and truthful is neccesary in this scenario
-
Cybersecurity Advantages
Examples of Advantages
-
Spotting patterns, variations in patterns, offering early warning signs, compromised accounts and unauthorized attemps
Examples of Scenarios
You work for a company that is getting a new AI system. You have a part in writing the terms and condtions and your boss directs you to leave out some of the parts about data collection. He says that the cybersecurity advantages of AI will ensure that consumer data is not at risk. Should you listen to directions or refuse to lie to consumers.
The ethical decision in this scenario would be to refuse to leave out some data collection information in the terms in conditions. AI systems are not perfect and there are still potentials of risks. Consumers have the right to be aware and consenting.
Rights- moral authority is determined by indivdual rights and each consumer should be fully aware of the true terms and conditions, then have the right to consent or not.
The unethical decision would be to go along with what your boss in saying and disgregard ethics/morals
You work for a company that utilizes AI. As you are about to finish your shift, a threat is detected by the system. Recently, the system has been flagging things that turn out not to be threats. Should the threat be ignored or investigated?
The ethical decision would be to investigate the treat, even if you believe it may be faulty
Utilitarianism- Just by checking to ensure data is not at risk, the greatest amount of people will be benifited and protected
-
Ethical Challenges
Types Of Challenges
Bias and Discrimination
Biases programed can cause imbalences, and discriminatory effects
Lack of Transparency
It can be difficult to understand how they make decisions and forecast future and accountability issues result from this lack of openness
-
-
-
-
-
Examples of Scenarios
An AI programer who was once robbed by an African American man, has to decide whether to include racial bias into his AI system
The ethical decision would be to avoid the use of personal bias as it can have discriminatory effects such as racial profiling
Consequentialism Theory-an action is right or wrong based on its consequences. In this case, the AI system would would not have an unfair and unjust bias against all African Americans
-
A sucessful business owner has to decide wether or not to fire a large amount of employees due to the potential of AI taking the place of their jobs
The ethical decision would be to not replace human jobs and intelligence with an AI system. Although the system can be utilized, jobs should not be jepordized and taken away
Traditional Ethics such as Confucian theories- What you do not desire for yourself, do not do to others
-
-