Artificial Intelligence (CO3519) Tutorial 1.3: Discussion

For the problems, see tutorial document 1.3.

As is generally the case in working on decision making and decision support scenarios, there are no right or wrong solutions; below, some examples are included to guide our discussions.

1.3.1 Objective spaces

Note that there are different ways of making sense of the acronym "SMART." To Doran (1981), it meant:

• Specific - target a specific area for improvement;
• Measurable - quantify or at least suggest an indicator of progress;
• Assignable - specify who will do it;
• Realistic - specify what results can realistically be achieved, given available resources;
• Time-related - specify when the result(s) can be achieved.

For the objectives below, submitted by Tauheed Mir, Jamie Pickering, and Jamie Liddiard, SMART was understood to mean specific, measurable, achievable, relevant, and time-bound, which is one out of many interpretations that have been given to "SMART;" this means that (A) takes the role of being realistic, while assignability is moved to (M) and/or (R), and the overall meaning is shifted slightly as well.

Achieving Together

• S - Increase the quality of student work by having more group projects for the next academic year.
• M - The overall grades and quality of group work is higher than individual work with the same time until hand-in over the course of a test year. Decided by the assignment marker.
• A - Quality of student work increases. Student marks go up an average of 10% in each assignment.
• R - When working together, students can reinforce one another to improve each other’s work. This can help people learn more about the topic and do better than they otherwise would do.
• T - Judged on a per-academic-year basis.

Deduced KPI: Student's average mark increase; to be determined every assignment per academic year (target: all > 10%).

Being Proud

• S - Students and staff are confident that the work and effort they put in is worth the grade they are given.
• M - The students achieve higher grades for their work, 2-1s and firsts for the majority.
• A - Despite harder work, the staff can give students the knowledge base required to achieve higher marks and top-grade boundaries.
• R - By having higher average grades for students, as well as students that are prouder of their work, the university will be seen in a much more positive light by students and brings up morale, leading to more.
• T - Per exam/assignment in a module with a larger review at the end of an academic year.

Deduced KPI: Student's average grades per assignment; to be determined per assignment per academic year (target: all > 60%).

Remark: The two KPIs above might be fused into one.

Creating Opportunity

• S - Planning modules to incorporate industry opportunities and extra-curricular materials that help boost student progression.
• M - At least 2 industry talks and other opportunities not on the base curriculum per semester.
• A - Students are given these opportunities through the university’s resources and connections.
• R - Number of students choosing to attend is recorded and engagement throughout the events are prioritised.
• T - Monitored, per module, per semester by a module supervisor and student feedback.

Deduced KPI: Number of extra-curricular activities/opportunities available; to be determined per module within a semester (target: all > 2).

Supporting All

• S - There are no patterns found from grade comparisons beyond the tutors that taught or the amount of effort given by the student.
• M - Students are free to file complaints over cases of inequality or mistreatment to a student hub.
• A - Success measured by a lack of unsolved complaints, indicating it is working well.
• R - Ensuring on a yearly basis, no major incidents or complaints arise from the tutors or students that are associated with the institution.
• T - Reviewed every year.

Deduced KPI: Major complaints received by student hub; to be recorded each semester (target: all = 0).

1.3.2 Parameter spaces

The following scenario was contributed by Katib Hussain (slightly edited):

The university has considered to upgrade workstations over the Christmas holidays (2 weeks) workstations. For this, a minimum investment is needed (presumably, into equipment and staff), and a minimum number of staff need to be allocated to do the upgrade. Above these minimum values, the university has the choice to allocate additional resources. Up to 600 workstations would need to be upgraded, but for every 10 workstations, a day's maintenance is required.

A large maintenance team with too little investment would reduce productivity and reduce the amount of workstation upgrades. Similarly, too much investment on a small maintenance team would limit the use that can be made of the acquired equipment. A balance is needed to ensure enough workstations are upgraded for the amount of people that are working on them.

Parameters:

0. Investment x₀ (including labour cost), in units of GBP;
1. Number of employees x₁ allocated to the work.

Objectives:

The target consists in upgrading as many workstations as possible with a minimum investment. This can be given the following formulation in terms of two minimization objectives:

0. Investment y₀: Note that y₀ = x₀;
1. number of workstations not upgraded, out of the total 600 that require an upgrade.