CS 4580/5580 - Paper

As the syllabus states, you are required to write a short paper for this course, which counts for 20% of your grade. This is that paper.

Your paper should be around 5 pages (1500 words). You should submit it in canvas, as an uploaded file in word or pdf format.

For the paper, you should read the book Goal-based Decision Making.(GBDM) Stephen Slade. Hardcover: 304 pages. Publisher: Psychology Press (October 1, 1993). It may also be available at the Yale Bookstore. (I have not looked.) Online copy through Yale library Online copy of thesis from which book was derived at Yale Library

In so far as possible, you should incorporate ideas from the book into your paper. Note: you do not have to agree with the book. You can challenge or dispute it.

The Big Picture

Once you leave academia, you will find that the world is more subjective. You will face decisions that are not so black and white. The security analysis homework is an example. So is this paper. I am not looking for a right answer. I am looking for an informed analysis and cogent explanation.

Also, the GBDM framework provides a method for approaching ill-defined questions for which there are multiple, subjective perspectives.

Topics

• The Turing Test. Take a stand. Is it good or bad as a measure of artificial intelligence? Why?

  

• Searle and the Chinese Room. This paper originally appeared in Behavioral and Brain Sciences, a peer commentary journal. It was published along with a couple of dozen critiques. Write your own critique.
• Cognitive Modelling: Benefit or Necessity? Discuss the benefits and needs for AI programs to mirror human cognitive processes.
• Chatting up ChatGPT, Gemini, et al.. Siri and Alexa primarily respond to questions. What would it take to have a long conversation with a computer? I encourage you to take the cognitive modelling perspective and explore what it takes for a person to engage in a long conversation. What does ChatGPT tell us about human cognition?
• Driverless cars: the next frontier. As suggested in the video trailer for this course, the Trolley Problem is not the best test for driverless vehicles. What should a driverless car know in order to be more like a human chauffeur?
• Risk management. In the fintech world, risk management is the practice of mitigating the downside risk of your investments. That is, to minimize how much money you lose. As we learned, risk is often quantified by the volatility of your portfolio. In life, there are predictable risks which we can mitigate. In driving a car, you can get a flat tire. You can mitigate this risk by carrying a spare or belonging to a roadside assistance plan such as AAA. You have the risk of getting sick. You mitigate this risk in part by having health insurance (and getting vaccinated). In computer programming, we are also concerned with mitigating risk. In programming, risk is often equated with errors or exceptions. That is something happens which causes the program to break. Many of these exceptional conditions are known in advance, such as division by zero or missing files. Python has an extensive list of built-in exceptions. For this topic, I ask you to come up with a similar list of exceptions for finance and investing, preferably with risk mitigation plans. Examples include foreign exchange risk and credit risk.
• Emotions. What are the benefits of a computer program understanding emotions? How about a computer having emotions? Propose how to do this. See GBDM pages 95-97, 104.
• Experience. There is a saying: good judgment comes from experience and experience comes from bad judgment. Why do many people become less idealistic over time? Possible explanations include finding counterexamples to expectations, adopting conflicting goals through relationships, making bad decisions that need to be justified. Try to frame your discussion using the goal-based model of interpersonal relationships. Thus, how would a computer change over time through its experience and relationships?
• Anthropic vs. The Pentagon. As you may know, last week Anthropic refused to allow the Pentagon to use Claude for (a) massive civilian surveillance, or (b) autonomous lethal weapons. What do you think? (Note: OpenAi quickly raised its hand to volunteer their services.)
• Digital Ethics The five principles of digital ethics are beneficence, nonmaleficence, autonomy, justice, and explicability. In this class, we have emphasized the last point, namely that computers should not be black boxes. They should be able to explain themselves. GBDM demonstrates one framework for achieving explicability. How can you use the GBDM approach to address the other principles: doing good, not doing bad, deciding to decide, and societal good. Can you create an ethical computer? The GARP document ethics.html gives a very good overview of ethics in general and AI ethics in particular. Should you write about ethics, you may take the content of this paper as axiomatic. That is, you should not repeat any of this material, but may build on top of it. I would welcome a paper in which you use the GBDM framework to propose implementations to address one or more of the ethical issues raised in the GARP document. You get an automatic upgrade if you tackle this question using that approach. Also, you might consider implementing your ideas for the final project in the course. That is, write a paper describing how to implement an ethical program, then implement it.
• Dealer's Choice. If you have a different idea for a paper, you should run it by me. I am likely to approve it. If you are going this route, you should get approval before spring break.

ChatGPT / Gemini etc.

How to Write a Paper

• An introductory paragraph concluding with a thesis statement. The thesis statement is what you are going to demonstrate. It is the point of the paper.
• Three or more paragraphs or sections supporting your thesis statement.
• A concluding paragraph that recapitulates your thesis statement.

Most of you are friendly with Mr. Number and, by extension, mathematical proofs. An essay is like a proof. You begin a proof with a statement of what you intend to prove. This is like the thesis statement. Your reader should never be confused or uncertain about the point of your argument.

The steps of the proof are axioms or logical inferences that form the foundation of your argument. These are like the supporting paragraphs of your paper.

A proof ends with the conclusion, which is the original statement, albeit now you have proven it. QED stands for quod erat demonstrandum, "Which was to be demonstrated." Sometimes you may also see W⁵, or "which was what we wanted."

Your reader should always know what you are trying to prove.