Description

1 QUESTION 1 – COMPUTER SCIENCE PROMPT
1.1 Positive effect of OMSCS
The main reason for me and many others to choose OMSCS is because of the value. The program has a great reputation with an affordable cost. There are many low-cost online programs, but the quality is also an important factor. From my point of view, the main positive effect of OMSCS is to make high quality post-graduate education more accessible like never before.

Figure 1—Gatech OMSCS – Enable more people to access high quality postgraduate education
1.2 Potential Negative effect of OMSCS
Every coin has 2 sides. The low-cost structure makes the OMSCS program more accessible, but people tend to have lower commitment as well. The drop-out rate of the OMSCS program is higher than the on-campus program, one of the reason is that the sunk cost is much lower. People can afford to drop out with minimum financial cost. But there can be selection bias here as well. The low-cost structure enables much more people from lower-income families to be participated into the program, but those people are much more likely to drop-out because of the family issue. Also, another important factor for the high drop-out rate is that, the OMSCS students are generally working adults with full-time jobs. There are much more life commitments comparing to the normal colleague students.
There are many people dropped-out from the OMSCS program because of the change of work and family commitment (e.g. Having a new-born to take care).
1.3 Design to preserve the positive effect and limit the negative effect
To keep the program accessible with low-cost and encourage the commitment and engagement, we can increase the cost of the program and give as much scholarship as possible for those who are in needs. There can be performance based awards as well. To encourage people with major life changes able to continue the education, we can allow people to pause the program for 6-12 months
with valid reason.
2 QUESTION 2 – COMPUTER SCIENCE PROMPT
2.1 Area that political motivations are determining the design of technology
Social media keeps people connected nowadays. There are many active designs and studies in this domain. The way people sharing and interacting with each other are heavily affected by the political motivations. In China, the biggest social media is not Facebook or Instagram, it is the Moments within the super app WeChat. In this section, we will discuss the value sensitive design of WeChat Moments and stakeholders with their motivations. The way WeChat Moments allows user to share their life and how they are interacting is very different from what you can do with Facebook or Instagram.
2.2 Stakeholders and their motivations
We can refer to the below Table 1 for the stakeholders and their motivations. There are clear conflicts between the users and the advertisers, the users want less annoying ads and the advertisers want the opposite. Tencent (WeChat company) would love to maximize the profitability but if user experience damages too much, they could lose users to competitors. And the company can be in trouble with regulators if they misuse the user data. So they have to carefully design the product that according to the Chinese culture’s value and minimize the negative impact by monetization.
Table 1—WeChat Moments – Stakeholders and their motivations
Stakeholders Motivations
WeChat Company High user engagement, high user activities to make ads slots more valuable, profitability
WeChat Users User privacy is concerned, free to post without judgement, interaction with friends, less ads
Advertisers High user activities, more people to see their ads
Regulators User data is protected, users are free from social media scams
Competitors Focus on the bad experience part of WeChat Moments and offer a better
product
2.3 Motivations affecting the design of the technology

Figure 2—WeChat Moments – Value sensitive design
From the Figure 1 we can see a particular post of WeChat Moments. The core value of the WeChat design it to take care of the user privacy. It is very different from another social media platform called Weibo, which is like Twitter. Everyone can see others’ posts. The post can only be seen by the friends added. Also, you can set up group view access to allow only a small portion of people see your posts. Through this design, WeChat emphasize the privacy of the users and let them to choose who can see their posts. This is tailored to the WeChat’s motivation, because they want to have more user engagements, and encourage users to do more posting. The ultimate goal is to boost the engagement that they can sell other services like WeChat pay, and more advertising on the platform. The design concerns about the privacy also lowers the barriers for users to do posting, they can share some sensitive posts with close friends only.
3 QUESTION 3 – ACM CHI
3.1 Paper 1
“Wrex: A Unified Programming-by-Example Interaction for Synthesizing Readable Code for Data Scientists”
Authors:
“Ian Drosos, Titus Barik, Philip J. Guo1, Robert DeLine, Sumit Gulwani”
3.1.2 Summary
This paper proposes a new programming-by-example way for data scientist to do data wrangling. Traditionally, data scientist has a variety of data wrangling tools to use, but they are reluctant to do so. Because to use those tools requires data scientist to leave their programming environment – Jupyter notebook, and often in another programming language than Python. The authors present a new tool embedded in the Jupyter notebook using the same python language, which will make it easier for data scientist to adopt the new tool to the workflows.
3.1.3 Interesting point
First, as a 6 years data scientist, I agree with the author that the data wrangling tools are generally ignored by the professional data scientists. We normally have to write code with libraries like Numpy, Pandas and Matplotlib to understand how to do the proper data wrangling. But if there is a widget in the jupyter notebook to speed up this process, it is definitely going to be helpful. The authors

Figure 3—WREX – Jupyter notebook extension to do data wrangling
used the familiar interface to bridge the gulf of execution for the end users.
Second, authors have done enough research on the needfindings. There are many tasks for a data scientist to deliver a data product. Data wrangling is the most time consuming one but less interesting. This is a great area to design a product for improving the workflow efficiency. The authors also did both qualitative research and quantitative research to make sure the tool is indeed improving the data scientist’s efficiency.
3.2 Paper 2
“If I Hear You Correctly: Building and Evaluating Interview Chatbots with Active Listening Skills”
Authors:
“Ziang Xiao, Michelle X. Zhou, Wenxi Chen, Huahai Yang, Changyan
Chi”
3.2.2 Summary
This paper is written by joint effort of UIUC and Juji. It proposed an activelistening chatbot for interview process. Current chatbot is usually not good at handling free context out of its training domain. Especially for interview chatbot, it is more functioning as a better way to do form filling. But by leveraging the latest progress in natural language processing and chatbot technology, Juji is able to build a chatbot to handle users’ free-text response and open-ended questions. This is a great progress to enhance the user experience when they are interacting with chatbot.

Figure 4—A screenshot of an example interview conducted by a chatbot (AI Minion) and a user (Sara).
3.2.3 Interesting point
4 QUESTION 4 – OTHER CONFERENCES
4.1 Conference 1 – International Conference on Human-Robot Interaction
This is one of the best full paper awards of International Conference on Human-
4.1.1 Title and Author Paper Title:
“Explainable AI for Robot Failures: Generating Explanations that Improve User Assistance in Fault Recovery”
Authors:
“Devleena Das, Siddhartha Banerjee, Sonia Chernova”
4.1.2 Summary
In this paper, the author proposed a new way of interpreting error message using explainable for the novice users. The author proposed a new type of explanation, Eerr, to help the non-expert users understand robot failure and identify solutions. By using the explainable AI approach, they are able to generate meaningful explanations for unseen scenarios as well. This is very helpful for the non-expert users to understand the problems of the robot and act accordingly to fix the issue.
4.1.3 Interesting point
The authors have creatively to adopt the explainable AI methods to robot failures. It is one of the hottest areas in machine learning field and it make sense to adopt some best practices to the robotic field. Especially, the failures of robot is almost inevitable during the development phase and moving from testing environment to the real world.
4.2 Conference 2 – International Conference on Tangible, Embedded, and Embodied Interaction
4.2.1 Title and Author Paper Title:
“ForceStamps: Fiducial Markers for Pressure-sensitive Touch Surfaces to Support Rapid Prototyping of Physical Control Interfaces”
Authors:
“Changyo Han, Katsufumi Matsui, Takeshi Naemura”
4.2.2 Summary
The authors proposed ForceStamps, fducial markers for supporting rapid prototyping of physical control interfaces on pressure-sensitive touch surfaces. The authors also showcased a wide range of physical controls could be prototyped by utilizing the characteristics of the buffer materials and the spatial constraints.
4.2.3 Interesting point
We rarely understand how to design and rapid prototyping for the haptic sense. It is an interesting and useful design that enable designers without knowledge of electronics to be able to design for touch interface. ForceStamps bridges the gulfs of execution and gulfs of evaluation between the user and the design task. Also, from the example we can see that, it allows the designer to directly manipulate through the interface. This interface further shorten the path between the user and the task.
5 REFERENCES

Figure 5—ForceStamps – Direct Manipulation
national Conference on Human-Robot Interaction. HRI ’21. Boulder, CO, USA:
Association for Computing Machinery, pp. 351–360. isbn: 9781450382892. doi: 10.1145/3434073.3444657. url: https://doi.org/10.1145/3434073.
3444657.
CHI Conference on Human Factors in Computing Systems. New York, NY, USA:
Association for Computing Machinery, pp. 1–12. isbn: 9781450367080. url: https://doi.org/10.1145/3313831.3376442.
//doi.org/10.1145/3313831.3376131.