# My fourth week at AIMS

After having it fairly easy during our first 3-week long block at AIMS, the start of the second block has been quite the shock to my system this week. The two courses we are doing during this block is Physics Problem-Solving and Statistics.

I have never done much statistics other than the bare basics, and I haven't reviewed physics since 2014 - more than 10 years ago now. I also don't have a great long-term memory, so I'm needing to relearn a lot of the basics of calculus and physics very quickly.

The lecturers are moving fast and assuming that we know the basics. At times I feel very overwhelmed by the prospect of catching up, but then I remind myself that that is why I came here: to have the motivation of needing to relearn and revise my mathematical knowledge.

I've also gotten sick this week and I've been battling to keep my Autism-related sensory overload under control. Being in a classroom and not having control over the temperature, sounds, smells, textures, etc. is very different to the work-from-home environment I've had for the last few years.

All in all, however, I'm still very glad that I entered the programme at AIMS. It feels like a huge privilege and it's challenging me in a very satisfying way.

It has also made me think a lot about the future of AI and our society, partly because the other major programme at AIMS is a programme in AI for Science for Master's. But also because of all of the news around DeepSeek the past two weeks.

My fellow students and I have also been using LLMs extensively to assist us in our learning. I have used it to explain concepts to me that I didn't understand and to generate LaTeX and R code for me, because I don't want to spend hours on learning these two languages.

I'm trying to only use LLMs to supplement my knowledge, but since the reasoning capabilities of LLMs are becoming increasingly better, one could definitely use it to assist you with your assignments. I can see how someone could even pass our courses by using only LLMs for their assignments. Lecturers will need to consider this in the future when doing assessments. Unfortunately, the next best assessment option I can see is good old closed-book tests.

I have subscribed to a newsletter for post-graduate opportunities and it seems like the majority of new opportunities incorporate research around AI. At this stage, I don't want to become a researcher in AI specifically, but I know that I will need to become knowledgeable in utilising AI in whichever field I do end up entering.

However, for the next two weeks, I will be spending my time learning statistics and physics. Before 9 PM tonight, I need to submit my physics assignment and I'm still trying to understand the last question about harmonic oscillators.