✨ The AI Quirk: You give AI a clear instruction at the start of a prompt… but the response seems to ignore it completely. Even stranger, if you repeat the instruction later in the prompt, suddenly the AI follows it perfectly. ✨ What’s Going On: - Large language models weigh instructions "based on proximity and clarity" within the prompt. - Instructions buried early in a long message can lose influence once the model begins predicting the response. - The model often prioritizes "the most recent instruction signals" it sees. - If a prompt contains mixed signals (examples, context, and instructions together), the model may treat the first instruction as "background instead of a rule". Example: You start with: 1) Write this in bullet points. 2) Then provide a long paragraph of context. The model may treat the context as the main task and default to paragraphs. But if you end the prompt with: “Use bullet points for the final answer”, the output suddenly follows the rule. ✨ What To Do If You See It: - Place "critical instructions at the end of the prompt". - Separate instructions from context using spacing or labels. - Repeat important constraints when precision matters. Try this prompt: “Using the context above, produce the final answer in bullet points only.” ✨ Why This Happens: AI isn’t reading instructions like a human would. It’s predicting the next most likely text — and "AI tends to pay the most attention to the instructions it sees last." ✨ AI Bits & Pieces — helping people and businesses adopt AI with confidence.

We were at dinner with a friend, @Mark Zayec, for his birthday. As we were exchanging AI stories he started telling us about an interaction he had with ChatGPT. For the past year, he’s been speaking small amounts of Italian and French into it — mostly to help himself learn. He’ll throw in things like, “Buongiornata mio fratello 🇮🇹.” On that day, ChatGPT responded in such a way he needed help with the interpretation? Therefore, he cut and pasted it into Google Translate to interpret it?” 🤔 I said, “Wait… so you spoke to ChatGPT in Italian instead of English, it responded in Italian… and then you needed it interpreted?” “And then you cut and pasted it into Google Translate to interpret it?” He chuckled, and said “yes.” Anticipating what was coming next. Without even thinking, I said, “Why didn’t you just ask it to interpret it in plain English?” We all looked at each other, and busted out laughing. 😂 This was a perfect illustration as to how we are still wired to think tool-to-tool instead of conversation-to-conversation. Even when we’re already inside the interface, our instinct is to jump somewhere else instead of just continuing the dialogue. You can say: “Translate that.” “Explain that in English.” “Rewrite that more simply.” It’s not about perfect prompting. Or jumping to another app or tool. It’s about realizing you can just keep talking. That’s AI in Real Life. Note: Animated comic created with Nano Banana 2.

We were deep in conversation, driving to Florida. Talking at some length about the 2026 strategy for AI Bits & Pieces. What stays. What evolves. What new ideas might want a little room to grow. After a lot of back and forth, there was a pause. The kind that means someone is thinking, not finished. Then Michael started thinking out loud — as he does. I was listening. Or so I thought. Michael was reciting a finished thought — pulling together all the pieces of the conversation we’d just had about AI Bits & Pieces and its next chapter. Naturally, I answered. Quickly. Confidently. Like a spouse who’s been married a long time and knows the rhythm of these conversations. And then ChatGPT started talking. And then it just… stopped. Like, oh — sorry, go ahead. I remember thinking, "Why did it start talking?" Completely forgetting that Michael had ChatGPT set to voice mode to capture our thoughts and notes. So, I kept going. Added a little more context. And then, suddenly, ChatGPT jumped back in and essentially said, “Yes, I agree with Michele.” 😳 I looked at Michael with that "what just happened" face. That’s when it clicked. He wasn’t asking me. He was asking "TARS" (yes, from *Interstellar*) — as Michael calls ChatGPT. And somehow, without meaning to, I had jumped into a three-way conversation… and the AI wasn’t waiting at all — more like a cat behind the couch, ready to spring. I didn’t know whether to laugh or shake my head in bewilderment. Probably both 😂 And I thought to myself — "damn… it’s already here." Woven quietly into our conversations, our thinking, our planning. It made me wonder — where else is AI showing up that I’m not even consciously aware of? We’re just going on about our day — the kind of conversation I’ve had a thousand times with my partner of 27 years — and it’s already inserting itself into our lives. And maybe that’s how the biggest changes arrive — already settled in, before we realize we’ve adjusted. And that's AI in Real Life...

I’m excited to share something special with our AI Bits & Pieces community, especially the AI-curious members of our community. 🥁 🚀 Starting this week, my wife Michele will be contributing a new series called AI in Real Life — a warm, honest look at what it actually feels like to learn and use AI from the perspective of an everyday, real-world user. ✨ AI in Real Life is for anyone who’s ever thought: “I’m curious… but where do I begin?” This series will follow her personal journey with ChatGPT and other AI tools as she explores how they show up in everyday life — conversations with family, planning and organizing, trying new ideas, and even navigating the hesitation many of us felt in the beginning. Each week, Michele will share a short story, a small discovery, or a real-life moment that brought AI into her world in a simple, human way. My hope is that her voice helps make this community feel even more welcoming for those who are just getting started. ________ 🕰️ By way of background, Michele (@Michele Wacht ) spent twenty years as an executive selling services to the automotive OEM industry. She came from a corporate marketing and sales background, achieving top salesperson status at her company for many years before stepping away eight years ago to prioritize our family and be fully present for our daughter during her teen years. Now that Emma is off to college and recently turned 21, Michele felt ready to re-engage. And to my delight, she decided to join AI Bits and Pieces in helping people understand the benefits of AI — not from the perspective of an engineer or a strategist, but from the vantage point of someone discovering her own curiosity and how AI fits in as a life skill. If you’ve read Michele’s writing — as I and many of her friends have — you know she has a gift for turning simple moments into meaningful reflections. She approaches AI the same way — with curiosity, humor, and a down-to-earth honesty that reminds us that learning something new doesn’t always start with confidence. Sometimes it starts with dinner plans for friends, a college-age daughter on speed dial, and a willingness to try (a preview of her first post).

I had a moment this week that genuinely caught me off guard. I took a piece of writing I wasn’t happy with, moved it over to Claude expecting a softer, more polished take… and instead got something I didn’t expect at all. Claude swore. This video leans into the funny side of AI — the moments that catch you off guard and make you laugh. At the same time, it hints at something bigger: how quickly these systems are developing tone, personality, and presence. Curious if you’ve seen moments like this too.