Austin S.

8/20 - 8/22: First week of school and I got into WISRD. Already started working on building a computer to run augmented reality stuff.

8/25 - 8/29: Second week in WISRD, we took inventory of our computers and their specs. I worked with the REAL XR lab to construct sensory-distorting glasses and studied their effects on vision. We are also considering building a personal storage server with the spare parts we have.

9/2 - 9/5: Third week. I started researching AI and machine learning to make a prototype face recogniser AI. I also went on a trip to the Ballona Wetlands to collect some water samples and data on the spread of native plants within the wetlands.

9/8 - 9/12: I continued researching AI and the details of what actually happens when it receives a prompt. I did this so I could expand my knowledge of AI, so I can be on the same page as other people working on this project, so we can collaborate effectively and get more done.

9/15 - 9/19: Our lab decided on a project to work on. We decided to research how AI scrapes websites, what it looks at, and what it doesn’t. I researched how websites are optimised to appear higher on search results. We also discussed the costs related to our possible projects and whether they would be feasible.

9/22 - 9/26: I continued researching my lab project. I researched web hosting services and compared them to alternatives to see which service would be the best for our project. We evaluated them based on factors such as price, whether we paid monthly or by traffic volume, how much analytics we had access to, and how easy it is to use the service. I also developed a booth for Dark Matter Day with Leo. We decided we would demonstrate “hot ice”, where a solution of sodium acetate rapidly crystallizes when disturbed, heating up and forming crystals that look like ice. We sadly were not able to complete the experiment because we ran out of time in class.

9/29 - 10/3: We tested the new DeepSeek 3.1 Terminus model. We tested its speed and capabilities across different areas. In this process, we discovered an anomaly in the model where, after asking the prompt “whats the max amount of tokens you can use at once” to the non-reasoning model, we experienced degraded performance on subsequent prompts. In response to this prompt, the model also hallucinated and answered a non-existent prompt of “1+1=?”. We are currently unable to recreate this response.