PHDDS91823 Artificial Intelligence – Hypes and Hopes (Spring 2024)
Facts about the course
- ECTS Credits:
- 5
- Responsible department:
- Faculty of Computer Science, Engineering and Economics
- Campus:
- Halden and/or Fredrikstad
- Course Leader:
- Stefano Nichele
- Teaching language:
- English
- Duration:
- ½ year
The course is connected to the following study programs
Elective course in the PhD programme Digitalisation and Society.
PhD candidates from other institutions are welcome to apply, but students from this PhD programme will be prioritised.
Recommended requirements
Elementary knowledge of computer science and programming.
Basic knowledge of mathematics and statistics.
Lecture Semester
2nd and 4th semester (spring)
Spring 2024: March 11-13 and May 29-31
The student's learning outcomes after completing the course
Knowledge
The student has
-
an overview over important terms and concepts within AI
-
knowledge of ethical issues involved when developing and using AI systems
-
knowledge about practical challenges with AI systems, including overfitting/underfitting
-
an overview over techniques available to solve ethical issues when using AI systems and their limitations
Skills
The student can
-
choose the correct algorithm for a given type of data
-
empirically evaluate AI systems
-
discuss and address ethical issues involved when collecting data for, developing, and using AI systems
-
develop simple AI systems using basic machine learning algorithms
General competence
The student has improved his/her competence in
-
treating and analysing data
-
evaluating AI systems realistically
Content
Artificial Intelligence (AI) has become much-discussed topic in most academic disciplines and general mainstream news and media. The promise of AI is seemingly great, as it can purportedly help us solve just about all scientific problems, enable autonomous transportation, lead to new business insight and innovation, help improve political processes, fight pandemics, help reach the sustainable development goals, and so on. However, other voices argue that AI is associated with a number of important flaws and shortcomings. Some focus on how the capability of AI to solve problems are exaggerated, and others focus on the importance of considering the potential negative consequences related to the process of developing and using AI systems – whose behavior might not be fully understood. This course aims to provide the students with a deeper understanding of what AI really is, and its true potential, allowing them to evaluate, recognize and distinguish between AI hype and hopes based on a realistic assessment and appropriate use of AI.
A lot of different terms are associated with the concept of AI, such as narrow and general intelligence, supervised and unsupervised learning, neural networks, and deep learning. These terms are regularly used in research articles without being defined, and in this course we will unpack and explain them, along with the concept of AI in general. The students will learn how to use basic programming and statistics skills to develop and evaluate basic AI systems. This will both allow them to choose appropriate models and applications, and they will be able to understand and engage with research involving AI algorithms in their own and other fields.
As the students learn to realistically assess the limits of AI systems, they will also learn of the dangers associated with what these systems can be used for. Because the logic behind AI systems' decisions can be difficult or impossible to follow, the use of AI systems presents unique ethical challenges involving, for example, fairness and bias, accountability, transparency, and issues related to privacy. The status and potential for explainable AI will thus be emphasized.
Privacy is another key topic in this course, as many important uses of AI systems involve the collection and processing of personal data. This course aims to provide the students with an understanding of how sensitive data can be protected using data anonymization techniques and privacy-preserving models. Furthermore, legal frameworks, such as GDPR, controls how AI systems can be used in the EU, and the students will learn how compliance with such frameworks is handled.
This course is suitable for everyone with the recommended previous knowledge, including experienced and inexperienced AI practitioners.
Forms of teaching and learning
Lectures, seminars and individual presentations, peer review of fellow student’s paper, and discussions.
The duration of the course is six days, distributed over two or more sessions.
Workload
The workload is estimated to 130 hours.
Coursework requirements - conditions for taking the exam
-
A minimum of 80% coursework attendance is required.
Examination
Individual assessment is based on two parts:
-
An individual essay/article of approx. 3000 words. The students choose their own topic related to the course content, which must be approved by the course convenors.
-
An individual oral presentation where the candidate presents the content of the essay/article. Duration approximately 20 minutes, followed by a 10-minute discussion.
The essay/article must be approved before the oral examination.
Grading scale: Pass/Fail
Examiners
One internal and one external examiner
Conditions for resit/rescheduled exams
Same requirements as the main exam.
Course evaluation
Feedback from our students is vital in order to develop and offer high quality courses. The course is evaluated using an oral evaluation conducted at the end of the course.