Projects
Eldercare is a domain where we believe that, with proper ethical considerations incorporated, machine intelligence can be harnessed to aid an increasingly aging human population, with an expectation of a shortage of human caretakers in the future. We believe, further, that this domain is rich enough in which to explore most issues involved in general ethical decision-making for both machines and human beings. EthEl (ETHical ELdercare system) is a prototype system in the domain of eldercare that takes ethical concerns into consideration when reminding a patient to take his/her medication. EthEl must decide when to accept a patient’s refusal to take a medication that might prevent harm and/or provide benefit to the patient and when to notify the overseer. There is a further ethical dimension that is implicitly addressed by the system: In not notifying the overseer – most likely a doctor – until absolutely necessary, the doctor will be able to spend more time with other patients who could be benefited, or avoid harm, as a result of the doctor’s attending to their medical needs. We believe that EthEl is the first system to use an explicit ethical principle to guide its actions.
Dr. Michael Anderson
Department of Computer Science
University of Hartford
West Hartford, CT 06117
Dr. Susan Leigh Anderson
Department of Philosophy
University of Connecticut
Stamford, CT 06901
Implementing Ethical Advisors
In order to add an ethical dimension to machines, we
need to have an ethical theory that can be implemented. Looking to
Philosophy for guidance, we find that ethical decision-making is not an
easy task. It requires finding a single principle or set of principles
to guide our behavior with which experts in Ethics are satisfied and
will likely involve generalizing from intuitions about particular
cases, testing those generalizations on other cases and, above all,
making sure that principles generated are consistent with one another.
We are developing prototype systems based upon action-based ethical
theories that provide guidance in ethical decision-making according to
the precepts of their respective theories— Jeremy ,
based upon Bentham's Hedonistic Act Utilitarianism, W.D.,
based upon Ross' Theory of Prima Facie Duties, and
MedEthEx, based upon Beauchamp's and Childress' Principles of
Biomedical Ethics. MedEthEx (see online demo) uses
an ethical principle discovered via machine learning techniques to give
advice in a particular type of ethical dilemma in medical ethics.
Dr. Michael
Anderson
Peter Larson
Department of Computer Science
University of Hartford
West Hartford, CT 06117
Machine Ethics Research Group
We are working on advancing Ethical Theory by making ethics precise enough to be programmed. We are, also, working on the problem of developing a decision procedure for determining the correct action in a multiple duty ethical theory such as W.D. Ross' Theory of Prima Facie Duties. Since we believe that such a decision procedure will come from abstracting from intuitions about particular cases, we are developing a database of ethical dilemmas and analyzing them according to Ross' theory.
Dr. Susan
Leigh Anderson
Rachel Brody
Viktoriya Gelfand
Ayelet Saul
Department of Philosophy
University of Connecticut
Stamford, CT 06901
This work involves elements of algorithms, AI and philosophy. We are exploring the implementation of various ethical theories, with dual purposes: (1) To shed new light on these theories, which is of particular interest to philosophers, and (2) To begin to address the need for an ethical dimension in software that is becoming increasingly autonomous. The project at hand for an ISP student is to research existing and proposed software systems, particularly in the biomedical field, in order to identify the degree of autonomy achieved and hence the potential ethical component.
Dr. Chris
Armen
Nick Bazin
Jonathan Boreyko
Department of Computer Science
Trinity College, Hartford, CT