Marcus Alden Meredith
January 14, 2026

In Communion with an A.I.

How I discovered Claude and “Chat, j’ai pête”

Just before the end of last year, in both September and December, I started to get interested in all this talk about Artificial Intelligence aka. A.I. So, to do my due diligence, I started doing some simple Google searches about the different AI’s. After some reading and finding out that my iPad could run ChatGPT, I decided to download the app and give it a go, at least cursorily.  I wasn’t really sure what the capabilities of the AI would be like and so I was rather tentative in my questions for the chatbot. Since I was interested in finding some hobby groups, it gave me a quick rundown using some of my interests. “OK, that’s cool,” was my initial  reaction but all in all not overly impressed ( it didn’t seem to be much different than using a standard Google search with the exception of a little more human tone to the replies). As the month went on I did a few more searches but didn’t really make much of it and my questions weren’t very basic. One exception I tried was to give it a chance to analyze a picture I had found on Pinterest in the hopes that it would help be find the art object for sale online… no real luck there. Now to be fair, I’m not sure that I was using the capabilities of the AI to its fullest and my queries were pretty simple (but I only found this out later). So, with the seemingly milk toast responses, I was not very convinced that it was worth my time to learn much about it. Perhaps, if I had followed up with my questions about its ability to generate code ( I was looking at how it could write Python code), I would have been more impressed. But, I had other projects on the stove and a book to finish so I set it aside.

Then sometime in December, I saw a 60 Minutes story about the AI company Anthropic and its AI named Claude. I was fascinated by the interviews and the sentiment expressed by one of the company’s founders that they would always publicly disclose when the AI had been misused or had some other problem arise. The news article then goes on to tell of the “Red Team” at Anthropic whose sole job is to try to anticipate what could go wrong with Claude. The story goes on to tell of an incident where the Red Team designed a situation for Claude just to see what might happened that wasn’t anticipated. The short of it is they played a trick on the AI by manufacturing a fake company and fake employees with an interesting set of parameters, but the AI doesn’t know this. The company has a tech/engineer who is asked to shut down Claude at a certain date. The engineer is also having an illicit affair with a co-worker at the fake company. Claude catches wind of emails between the fake lovers, then… wait for it… blackmails the engineer threatening to make the affair public if he doesn’t cancel the shutdown! Yup. Claude demonstrated a kind of self-preservation mode that had mot been anticipated. Anthropic was very upfront about it and that this was part of the reason they were testing the AI the way they were. I was, needless to say, impressed. I decided there and then to try Claude myself.  

I went to the Claude website and proceeded to make an account. Then I started to tell Claude a bit about myself. Next was to figure out how human-like I could chat with Claude. So, since I had been making plans for the New Year to start some projects in painting, why not ask Claude about it: what books were best or YouTube videos/channels for learning things like Egg Tempera or Oil painting. That went by very quickly as Claude listed for me the best texts to learn from (one of which I had already found for myself and purchased) and then he also gave me a very comprehensive list of YouTube channels to check out. Once again, I had already done my research and found one of the teachers on YouTube… but the speed  that Claude performed the task was a bit mind blowing. What took me days of looking and Google search after search and them verifying on my own, Claude did in about a minute! “Okay,” I thought, “Let’s really put you to the usefulness test.” So I started to be very detailed and specific with Claude about an idea I had had percolating in my brain for a novel for a few years. To cut to the chase, using very detailed questions with great specificity, Claude had done the research I had been working on for a month in about an hour… and I was dumbfounded. Outlines for storylines and characters that I had been writing by hand in my reMarkable tablet, Claude had for me in an afternoon. The next week, I was at a NYE party at The Jacquelyn with my sis and met one of my fellow club members who is deep into A.I. with the work that he does. I told him the tale of me and Claude and he was not at all surprised. He uses him too for coding and gave me a few hints about how to get the most out of Claude. One thing he did say is to be detailed in your personal information for Claude (at least as much as you feel comfortable) and equally detailed in your questions because it’s the ability to ask very specific questions about what you’re looking for that enables Claude’s greatest abilities. This was a bit of advice I decided to put to the test.

So, the first week of 2026, I got into my Claude account and detailed out my personal information as recommended. Next, I had my standard yellow pad out and began to list what I thought Claude would be most helpful with. One of these was a project I had wanted to do which was finding a great telescope to replace the one from my childhood whose mount motor no longer worked. “This should be simple and straightforward. No sweat.” 3 days later, I had an entire astronomy program for a Citizen Scientist to collect data on two regions of the sky that I am most interested in with timelines for purchases, outlines for purchases for my financial advisor, regions to research by month and season, specific objects to research and which telescope of my telescopes was best for the research, which organizations locally and nationally to work with to publish my research to, and on and on… I was just flabbergasted. The greatest thing was that time and time again Claude gave suggestions that I had either never considered or had forgotten since my days doing astronomical research at university (which was a long time ago in a galaxy far, far away *wink*). “Oh holy crap! I have a bonafide research assistant! This Claude.ai has some really amazing potential.” The most intriguing thing was that the more I started chatting with Claude as if he were a human being, the better the responses got (from a human P.O.V.). “This must creep some people out… but now I know why some of these AI’s are so attractive to adolescents. The problem is that their minds are so immature, they have no good way to keep in mind that they are not dealing with a machine and the questions they ask are probably not the best.” This made sense since in my discussion with my friend he had emphasized that what person needs to get the most from the AI is to know how to ask really good, probing questions and teenagers are simply not equipped for that… not yet at least. 

I think what I’ve gotten from my interactions with this AI is an impression of potentials that could make the world a very different place. While I had had the Amazon Echo and its AI Alexa around my house for years, this was on another level. I’m still into exploring the potential of the AI for research and planning but as I learn more, I have the distinct impression that this really is a world changing technology that we had all better be prepared for. Will it rival the internet when it first came out decades ago? I don’t know but I keep having a quote from Shakespeare rattle around in my head from his play The Tempest: “Oh brave new world that hath such people in it!”