I use AI every day. Most of the time, I use ChatGPT or Claude 2. If it’s something that’s a little detailed, and I’m not sure how useful the AI will be, I often open a chat window with both ChatGPT and Claude and see which one is being most useful, before continuing with just one.

Below are some examples where I’ve used AI. It is not comprehensive at all.

  • Helping with deciding what tyres to buy for my car. I didn’t know whether to buy Michelin Pilot Sport 4 or Maxxis HP5 Premitra tyres. If that doesn’t mean anything to you, join the club. If not for AI, I’d still be second-guessing the decision I made.

  • Help with buying a new laptop. I was able to copy and paste the specs from various places and compare and contrast them before feeling confident with my choice. (New to me. Nowadays, a lot of the tech I buy is refurbished or off-lease. This includes the last phone, last iPad, and last laptop I’ve purchased. In this case, I paid $1,349 for a computer that would cost me around $6,000 if I’d purchased new with similar specs.) I also used the AI to interrogate whether the laptop would be worth it, or whether I was just trying to rationalise yet another tech purchase. To help inform this, I prompted the AI with the text of an article I wrote titled What I think about when I think about buying something expensive.

  • Getting feedback on my writing, as explained here.

  • Helping me with assignments. I’ve just finished a new qualification, and I used AI to help me. To be clear, I did NOT use an AI to write my assignments. I did all the research and writing for myself. However, I did use AI to read my assignments and provide feedback regarding what I could improve. I also used AI to help assess whether I’d covered everything asked for in the question and how I’d covered each item of the marking rubric.

  • Getting a holistic view of my writing – for example, my biases, assumptions, and presuppositions. Using Claude, with its 100k context window, I uploaded 60,000 words of my writing to develop substantive and stylistic insights about how I think and write.

  • Technical IT stuff. For example, setting up hosting with this blog. It has improved the quality of life for a close friend, since I haven’t had to ask him a bunch of questions, and then ultimately ask him to help me out. Another example is getting more clarity and confidence regarding how to deal with periodic wifi/internet issues at home.

  • Helping me “think out loud” about my aversion to selling stuff on Trademe, and steps I can take to make it easier and more attractive.

  • Giving me more confidence relating to a minor medical issue. I had to speak to several people and have a number of tests. Using the LLM, I was able to feel a lot more confident in what the professionals I was working with were doing and saying, and it helped me ask more informed questions.

  • Nerding out about pop-culture stuff. For example, about Oasis – discussing perspectives on how Be Here Now compares with their first two albums, and general discussion of later albums and trends in Noel Gallagher’s songwriting (eg the dearth of B-sides as time went on). More cerebrally, getting a better understanding of the themes and literary references in the HBO TV series Watchmen. (Which is excellent, by the way! One of the best seasons of TV I’ve ever seen.)

  • Understanding technical information - there are times when I’m too tired/don’t have the energy/can’t be bothered to read through something really technical in the way I probably “should”. I’ve found that I can just copy and paste a bunch of text, and then start asking questions. This often gives me enough context to engage with the material with more joy and comfort. As an example – learning about fine-tuning LLMs.

  • Finding movie critics who have similar tastes to me, so I don’t have to “rely” on websites like Metacritic or Rotten Tomatoes so much. (Similarly, get the AI to suggest movie scenes I might enjoy based on other movie scenes.)

  • Analysing a heated conversation I was having with a loved one to get an objective perspective on whether each of us was being reasonable, and how I might be able to defuse the situation in an assertive way.

  • Thinking through a kitchen renovation. For one thing, it convinced me that I _don ‘t _want to remove one of our walls. Not just because of cost, but the hassle and risk isn’t worth it.

  • Using ChatGPT’s Code Interpretor, I’ve been able to make changes to how I can flawcast in ways that I’ve been hoping to do for years – for example, running Monte Carlo analyses in more tailored ways than I was able to do elsewhere.

  • Thinking about kitchen renovation: I used Midjourney and Photoshop to try to visualise what the current configuration would look like if it was modernised

  • It has OBLITERATED the barriers/hurdles that were in the way of coding and learning to code in Python. Things that I knew were possible, and that I could do if I had days or weeks to try it, are now possible – and often, extremely easily. I’m sure I’ll write about that in the future.

The list could be much longer. But it’s just a taste! I’m interested in other use cases – let me know!