Hello and welcome to the first edition of this newsletter, which will complement my website AIS Tech Notes and the accompanying YouTube channel!
I feel like we’re getting used to the wow factor of every frontier model from every other AI company these days. Smartphones took it over a decade to get to a point where nothing really impresses anyone anymore. AI did it in... less than a few years?
First, some background stuff since it’s the first one:
I’ve written what to expect here in the last post. I’m trying out a structure where I’ll write a bit about something from the last week that caught my attention and single out three more stories that are worth noting.
Stuff happens in the tech world at the speed of light these days, and I’m trying my best not to turn this into an RSS reader for you, so no link dump.
Also, as written on the main website’s AI Usage Policy, rest assured that every word you read here is written by me. AI helps me with brainstorming and research, but I take the time to formulate my thoughts and put them into words so that they are hopefully worth your time.
My Take
One thing that made me think
Google says schools shutting down access to AI would lose access to search, too
An interesting development from Google that caught my attention: Google said if schools turned off access to AI, they would lose access to Google search, too, reports CNN.
The ominous threat from a monopoly angle aside, I think there’s an upside to this.
Schools really need to wake up to the reality of the changing world. I can understand that students today are more likely to ‘outsource’ all their thinking to AI and, as a result, fail to develop their critical thinking abilities. That may be the reason schools are rushing to ‘ban’ AI.
But AI is a useful tool; it’s here, and schools should figure out a way to implement it into their coursework.
I can say that because I’ve actually resumed my studies now that I don’t have 9-5 pressure anymore, and I’m pleasantly surprised to see that some of my course teachers have adapted to the AI era.
How? Well, they teach students to use the lecture materials as a framework and then ask AI questions based upon them.
AI gives us the analysis, which we then have to read through, understand, and formulate our own opinion on.
Because the exam is on paper, and we won’t have access to AI, we have to use our brains to think even though we’re actively using AI during study sessions.
I think this is the perfect way classrooms can adapt to the AI era: teach students to use AI to brainstorm, see different viewpoints, argue and counter-argue to find flaws in logic, and at exam time, let them write things by hand.
It isn’t impossible. I found out Sweden has gone back to pens and papers, but that feels a bit too radical. Schools must find a hybrid approach — a balance — between the banning of AI and going all in.
Worth Noting
Three things that I found worth talking about
ChatGPT is taking away the best thing about its paid plan
I’ve been a happy ChatGPT Plus user ($20/month) for a long time. My primary use consists of a lot of research, brainstorming, PDF and other document analysis, and tons of web-search/research.
I’ve never run into any limits in over a year of my ChatGPT Plus use.
I tried Claude Pro recently (purely due to peer pressure) and I was immediately disappointed at how it counts usage. Apparently, even regular conversations, even if you don’t do web research, apparently count against your quote. I realize you can use an older model to lower your usage, but I really didn’t like it.
ChatGPT never counted anything from its Chat interface against my quota.
I think that’s the best thing about ChatGPT — its practically unlimited ability to chat.
And that’s ending.
OpenAI is changing how it all works, and apparently, sometime starting December 2026, even Chat messages (which, I assume, include web research, something I do many times daily) will count against your usage quota.
I know Claude is best for builders of all types — designers, developers, and motion graphics, and even hobbyist makers, but for everything else, as a general-purpose all-rounder, I still think ChatGPT is better.
I guess I’ll find out how much of that is changing come next month.
Apple wants Siri to do more, but it still lags behind others
Siri AI is finally on iPhones, iPads, and Macs. I’ve tried it, and I remain unimpressed.
I think it’s a good fallback option since it can now answer simple questions or pull up information from the web (albeit much slower than its competitors), but we’ve been able to do that with other AI for what feels like a decade.
So when I read that Apple wants to make Siri more agentic so you can make actual changes to your computer and soon to third-party apps, I’m skeptical.
I hate that Apple Silicon-equipped Macs have become the go-to for local LLM experts, yet Apple, the maker of Macs, mind you, remains far behind.
macOS Golden Gate actually ships with a small ~3b local LLM called Foundation Models, but Apple doesn’t really promote this, and its use remains restricted to the Terminal, so the average user probably can’t make heads or tails of this.
How long until Apple realizes they are still sitting on a goldmine and they should make better use of the local LLM capabilities of Mac? What’s stopping them?
Claude launches Fable 5.2, and the world is....unimpressed?
Anthropic just released Fable 5.2, its latest frontier model, and, surprise surprise, it’s its most capable model yet.
I’ve seen a few posts popping up here and there about people building grand things — like an entire FPS game in hours — with Fable 5.2, but I also remember when people built grand things with Fable the original or GPT Astra and the others.
I feel like we’re getting used to the wow factor of every frontier model from every other AI company these days. Smartphones took it over a decade to get to a point where nothing really impresses anyone anymore. AI did it in... less than a few years?
Of course, everyone talks about AGI as the ultimate goalpost, and every frontier model seems like one step closer to it. That’s until you realize that no one really knows what AGI means or what “achieving AGI” looks like.
Until then, the latest frontier model is just the latest and greatest, and some people seem to be building grandeur things at an astonishingly fast rate. Yet, more people are without work. Cost of living is at its fastest.
Field Notes
Notes from AIS Tech Notes.. get it?
To AI or not to AI, that was never the question
It took me a long time to plan the whole publishing ecosystem — YouTube, newsletter, website, and all related social media channels — and during this time, I’ve kept one thing in mind: I’ll only use AI for research, brainstorming, and automating any repeated workflows.
I’ll never ask AI for an opinion and pass it as my own or ask it to “flesh out” a topic for me so that my abstract thought becomes a “polished” essay. I have nothing against those who do this, but I, personally, would rather you read my messy copy.
I’m seeing this sentiment across the web, too. While AI makes production super easy (whether that’s production of code or design), stuff humans actually read, watch, and listen to are things people want to have come from other people.
I already regret not writing since I started in 2008, and so throughout all the instructions to ChatGPT and Claude, I kept one thing clear, the finished product should only consist of words written by me.
In a world where people have automated the whole production pipeline (AI research comes up with new topics, researches some more and writes a script, generates images and voiceover, then creates a compelling video, all without any human input or oversight), typing words one letter at a time feels like an enormous waste of time and energy.
But I’m willing to do it. It gives me joy. It gives me purpose. At least, in another year or ten, I wouldn’t be left regretting not doing more of what I truly enjoy: writing.
So, yeah, whether to use AI for my content creation was never really a question. It was always going to be me.
Non Tech
Life can’t be all tech, right?
My wife and I watched Brad Pitt’s latest survival movie, Heart of the Beast, and I didn’t know how or when the 90 minutes passed. Definitely recommend watching it.
Grand Theft Auto VI is finally in my console! After what feels like an eternity of waiting time, the GTA 6 launch date is nearing, and I downloaded the massive 210 GB game file last night onto my PlayStation 5.
I’m still disappointed that it’s going to run at 30 fps. I don’t think 30 fps would make GTA 6 unplayable, but it’s still gonna be an eye sore for a while.
I just hope a 60 fps update will be available sooner rather than later.
Your Take
You really didn’t think I wasn’t gonna let you have your say, did you?
If you’ve made it this far, thank you so much! Really! I’d love to know what you think about anything I’ve written about so far and the overall length of it.
As I said, it’s all me with no AI oversight. In the past, I’d deliberately use Grammarly and similar tools to optimize every sentence. But for this, I’m choosing to do minimal edits and keep the text as close to my original voice as possible.
Of course, an AI can make it far more concise and better, but if you wanted to read an AI output, you’d just ask an AI. You’re reading this, so I owe it to you to spend the time to write it myself.
If you enjoyed this, please share it with a friend, and tell me this:
Do the capabilities of frontier models still excite you? Or have they become passing news for you, too?