Tag: BrookeW – Laber Labs

Today we spent our work day testing the game that we contributed some level designs for, Laber Lab’s Zombies on Treadmills. You can play the demo on your web browser here! (Just a reminder that this is a demo and not reflective of the final product)

Shown in the picture above is a level that I designed, it is a stage in the shape of a dinosaur 😊

The premise of the game is that zombies have taken over, and you want to use the treadmills (the platforms with the arrows) to direct them on a path through walking in the directions of the arrows towards volcanoes (which kill zombies) and away from houses (which contain people, who you want to minimize the deaths of). You can use the WASD keys to move the camera, as well as zoom in and out using the trackpad. You can change the direction of the treadmills by clicking on them to create a path for the zombies to walk. The “Zombie Noncompliance” variable is the chance that zombies may not follow the direction of the treadmills, which you will need to account for. Once you have created a path you believe should work, you can click the “Go!” button to spawn the zombies and see everything play out.

We were the first people to be able to test the game, and gave feedback to the team at Laber Labs on various aspects of the demo, such as the artwork, gameplay mechanics, sound design, and controls. It was certainly a one of a kind experience that I am grateful for Laber Labs for giving me the opportunity to have a glimpse at the behind the scenes of developing a video game!

Today’s workday was also over a Zoom meeting. We learned how to make a regression model using R, a programming language that specializes in statistical computing! We brainstormed ideas for making a model of data, using patients as an example. One way would be to plot the x and y points and find the line of best fit and create a mathematical formula/function y = f(x), and another way would be to find similar patients and utilize their already-existing data and formulas.

In the afternoon, we listened to Jesse Clifton, a Ph.D. statistics student at NC State University, talk about the morals and ethics of using artificial intelligence. First we were introduced to recent significant breakthroughs in AI, such as AlphaZero (an AI that plays chess, shogi, and go) and GPT3 (an AI that translates a text input into code). We learned about the risks of using artificial intelligence, such as fairness and reducing bias, catastrophic outcomes due to caring/prioritizing about the wrong thing, disempowering humans, and getting into conflicts.

Today’s workday was over a Zoom meeting, in which Justin Weltz of Duke University taught us about various sampling methods and social networks. We learned about the pros and cons between random sampling and biased sampling, as well as how to apply sampling probabilities to specific mathematic formulas. Additionally, we learned more about reinforcement learning (which we had touched upon previously) and its relationship to artificial intelligence and machine learning. We also got to ask Mr. Weltz about his experience as a current graduate student doing research projects at Duke, such as what it’s like to work on research projects and how to set goals for these projects, advice for graduate school, and what topics in math and programming he would advise to learn if one is interested in pursuing a career in data science.

Today, Alex Cloud, a current NC State graduate student who works at Riot Games  (the studio behind video game titles such as League of Legends and Valorant) came in to teach us about academic statistical concepts such as the difference between estimands and estimators, reinforcement learning, and luck and skill in games such as Randochess(p). He also showed us DALL-E, an artificial intelligence software developed by OpenAI that uses AI to generate digital artworks based on a prompt that is typed as an input. We got to play around and test it and it was incredible to see how the AI was capable of stitching together realistic and believable images and how it was able to understand many different prompts (that were taught by training the AI using images associated with captions). Of course, there were some shortfalls. For example, DALL-E interprets the input “tree bark” as a visual of a dog barking at a tree instead of actual tree bark, as is shown in the demo video for the software.

Shown above is an example from DALL-E’s website about how to use the software, I unfortunately don’t have any of the resulting images for the prompts we tested, however, some of the prompts we gave were “A painting of a cat by Caravaggio”, “Mario graduating from middle school”, “An Impressionist painting of Jinx from League of Legends”, and “Mona Lisa but as Zelda”!

Today Dr. Laber gave us a presentation detailing how data analysis can be utilized to solve real-world problems such as identifying cases of human trafficking through the analysis of emojis used on websites, finding which location is the most profitable area to place a restaurant, and logically concluding when to administer certain medicines to patients with specific conditions. Additionally, we learned about the possible benefits and repercussions of personalization and discussed probability and the relationship (or lack thereof) between correlation and causation. We also brainstormed methods through which we could relate non-numerical data such as words or emojis into basic mathematical formulas.

Here’s a picture I took of the Duke Chapel, where we ate lunch nearby. Fun Fact: The style of architecture of the Duke Chapel is Collegiate Gothic, an architecture movement prominent in the late 19th century and early 20th century that was a revival of medieval Gothic architecture!

Today Dr. Laber introduced us to the concepts of data science, and we sprung into our first project: making a quiz! You can play the quizzes we used as a reference for this project here.

The basic idea is to have the user choose which answer option (between a notable statistician and a random object) fits the question prompt. The main purpose of this quiz is not to judge users’ accuracy but rather to share facts about the subjects of the answer choices.

The statistician we chose is David Blackwell, who we researched facts about in our brainstorming. We also worked with designers who work at Laber Labs to create artwork for our quiz.

After having lunch, we worked on designing levels for Laber Lab’s upcoming video game Zombies on Treadmills, a process involving coming up with ideas for the shape of the map and where to place spawn points for zombies, houses, volcanoes, and treadmills.