I promised to post my first quarterly review, and I’m now sharing it.
I’m doing this for a couple of selfish reasons. Firstly, it helps me to hold myself to account, and secondly, it should help me to develop and expand on the brief thoughts I used to answer the questions in my daily journal. What I hope you will get is some ideas on how to carry out your own reviews.
The reason I decided on a quarterly cadence is that there is a reasonable amount of time between reviews, to allow experimentation, while being regular enough to make adjustments should they be required.
My Quarterly Review Questions
I currently have five quarterly review questions which are part of my AI and PKM framework. These questions themselves will need to be reviewed regularly based on both personal experiences and the outcome of future research on the cognitive impact of AI and other technologies.
Where have I become too reliant on AI?
My biggest reliance on Artificial Intelligence at the moment is around my writing. I have had problems with my spelling and grammar since I was a child. Even learning to read was a struggle. While my spelling and grammar have improved over the years there is still plenty of room to improve further.
That is why I’m happy to let Claude help me edit this post, and as part of that process it notes any patterns in my writing that I can improve. From a recent review of that note, I have a checklist to use on my initial edit, which should help me to improve my writing.
I always try to read through the AI-edited post before I publish to make sure I’m happy with the changes, catching anything the AI has missed or any errors I made myself.
What new AI capabilities have emerged?
Since the start of the year, Claude have released two new models, Opus 4.6 and Sonnet 4.6. Sonnet works well for everyday tasks. Opus is the more expensive model, but ideal for more complex and challenging tasks such as reviewing the state of my Obsidian vault during each quarter.
What does the latest research say about AI’s cognitive impact?
The latest research that I have read seems to suggest that there is a cognitive risk with using Artificial Intelligence and that it must be used with consideration. It suggests it is okay to use AI for skills you already know. But you need to be careful when it is used with skills you want to learn or improve.
This seems to fit in with the ideas behind my red lines which you can read about in Introduction to my AI Knowledge Framework level 0
Which leads me nicely to the fourth question which asks which of these red lines I was tempted to cross? The answer is none.
Am I using AI as a starting point or an endpoint?
While using AI with my PKM, it is definitely seen as a starting point, so no concerns there at the moment. My area of concern is around AI and my content creation, specifically around the editing of my posts.
Firstly, the editing happens towards the end of the process and I don’t always have the time to review the post due to my self-imposed timeline which I need to keep for consistency.
But I would argue that I’m using the editing of my writing as a starting point. This is because the AI notes and tracks any patterns it discovers in my writing and uses those as a starting point for checklists.
I also acknowledge that the AI helping me with SEO and social media is more of an endpoint, but it is all verified by myself and it’s a side of publishing my post that I have little time or interest in yet, though it is something I need to consider to help prospective readers find my writing.
Conclusion
Overall, I’m happy with the outcome of my first quarterly review. But I do need to keep an eye on how I use generative AI when writing as there is a risk in that area of my AI use.
Why don’t you join our monthly newsletter and get a regular update on our latest content?
Further reading
- How I use Claude with Obsidian: My PKM Workflow (Pt2): Second part of my series on how I’m using AI with Obsidian
- Introduction to my AI Knowledge Framework level 0: Introduction to my AI knowledge Framework and Level 0 of my knowledge framework