Here I am at the end of the first “month” of curriculum: I have “month” in quotations because this is actually week six, but I’ve had a few resets along the way due to life getting in the way. The journey so far has been fun and I’m excited to keep moving forward. The coursework is interesting and I feel like I have learned a lot in the last month and know more about math and AI than I did previously. My wife thinks I’m crazy and doesn’t understand why I’m doing this or the end goal, but that’s ok. She says the same thing about my armored combat team. Let’s break down the first month of work.
Both functions and polynomials seem to have stuck. Going into this coursework, I honestly have not done a single bit of math since I graduated college and that was Calculus I. I had no idea how to do even the simplest things as I had failed my way up in math since the beginning of middle school. When starting this course and beginning the math, things came back and started to click with ease, for the most part. Both functions and polynomials are concepts that I grasp now and can do with ease when seeing them, when before I did not even know where to start. Exponents and logs are in a similar boat: I had no idea before and no I’m comfortable with them. I had no idea when to add, subtract, or multiple exponents or when each was needed based on how the problem was laid out – I do now. They are no longer scary and I feel pretty confident in my ability to solve any equation that contains them.
I would say after going at this for a month, my biggest skill upgrade would be understanding how to do the problems and equations that were handed to me in the last month as previously I would have had no clue. I literally had zero understand or recall of how to do any math as it has been 10+ years since my last attempts. I feel confident now, or at least more confident that I was previously in my ability to do what I didn’t think I could do in previous years.
I wouldn’t say I underestimated the amount of work: I know how much it would be going in and it’s honestly not that much if I sit down and focus. As mentioned previously, its life getting in the way, and something unexpected popping up during the day like an impromptu work meeting that pushes back my studying. I’m not worried about it going forward as this kind of stuff happens, but I will hope to push forward and keep pace with the curriculum, finishing month two at the end of July instead of a few weeks into August.
Going in to month two, I want to focus more on difficult problems and concepts and really drill into them until I fully understand them. Currently I’ve just been asking Gemini to explain and guide me through solving some of the harder problems, but I feel like doing it this way does not really make it stick. I need to use the guided learning on more than a handful of example and keep at it until I feel comfortable and confident solving the problems on my own.
Moving on to a related topic: the US government lifted the export control designation on Anthropic’s Fable and Mythos models earlier in the week. They have been made available again as of a few days ago, with subscription-based usage going until the 7th and after they will only be available via API. This is good news in general, and I’ve been exciting and building / fixing things all week. Fable has revised a ton of code that Opus 4.8 wrote and in doing so found multiple critical bugs / vulnerabilities that it has since remediated. I’m learned to make the best use of my time with Fable: I had it update the claude.md file to align the other models better and tweak their performance in a way that would cause less issues and false assumptions than previously happened. I’ve learned to use Fable as an orchestrator to oversee projects and have it use the other cheaper models to perform the actual tasks, with itself doing QA and oversight. I think going forward this is how I will plan to do things: use lower tier models for a week building and tweaking things, then spend an hour or so with Fable to review the code and everything that shipped that week. This will probably cost ~$25 a week (I hope) and I think it’s the best moving forward. I wouldn’t mind having it be the orchestrator as that would still be cheaper than just straight Fable for everything, but then all models would be going through the API billing which is expensive considering I have a Pro subscription.
So, there’s my month one recap. As I stated, I’m happy with where I’m at and the progress that I have made. I have a few learning things that I need to change going into month two but I do not see them being an issue, and if done correctly will be beneficial for the rest of the coursework. I am excited to continue with the second half of Algebra 2 and move on to the rest of the math coursework. I can’t wait until I actually get hands-on with some models and build from the ground up with Karpathy and actual see stuff working that I built. I will have to have patience though as building the foundation is the most important part.