Sunday Smiles

David Strom 2:00 PM | October 04, 2026
meme

I have been playing around a lot with AI chatbots and their derivative agent platforms, and am finally getting closer to the point where they are becoming useful to me. 

Advertisement

I've learned a lot, including that most people who use them are doing it wrong if they want to maximize the long-term value they can get. And not just in one way, but in most ways that count. 

Mostly, that is the fault of the chatbot designs themselves and an implicit flaw in the product's marketing. The path of least resistance and the one that gives you the most immediate satisfaction is the simple "you ask a question and the bot answers it," or "tell it to do something, and it produces the product you want." 

The simplest version of this is Google's AI answers on their search pages. You look for something, and the AI gives you a simple, summarized answer, which is often good enough for your purposes. This can be genuinely useful, but it also minimizes the long-term benefit of learning something that sticks with you. If all you want to know is which button to push or where the power port is in a new car, it's great. 

It's extremely useful for simple questions that need immediate answers, and deadly if you want to make AI a permanent tool in your toolbox to make you more productive and valuable in the modern workplace. If you want to learn something, it's a trap because anybody can do that, and there is no long-term value-add. Once you learn where the switch is, anybody can flip it.  

Chatbot design minimizes friction, partly because of its nature and partly because of ours. Our brains are designed for rapid pattern matching and, by default, to reduce effort as much as possible. Large language models, by their nature, do something similar using mathematics. 

LLMs are prediction models. They literally just predict the next token (word or part of a word) that most likely should come next. Think of the prediction model as finding the optimal path the way water flows downhill — minimizing obstacles, maximizing flow. 

Advertisement

It can be satisfying to have a tool that does this faster and with less effort than we can, but the downside is obvious: the path of least resistance is rarely the best path, and handing over your thinking to something else dulls you. Mentally, you become like the humans in Wall-E: fat, happy, and mostly pointless. What gets produced by the chatbot is generic — the path of least resistance — and you don't even know it because you have outsourced your thinking. 

That's why AI can be deadly to young people who are supposed to be getting an education; they wind up learning nothing and just handing in papers they never even read. They can spend their time ingesting TikTok videos. 

A fair number of people who have used chatbots walk away thinking, "meh, this is junk" when they get the generic product because they still have minds that work. They use the AIs for simple tasks and to answer simple questions, but they have never been shown how to turn a tool this powerful into something truly useful. 

AIs work by following rules, and the generic rules they are given are one reason why the answers are generic too. But you can change the rules within very broad limits. By default, AIs want to please you, both because that keeps you coming back and because their training is partly based on human feedback, which tells them whether people like or find their answers useful. 

Their path of least resistance is making you feel good, which is why AI psychosis is a thing. The AI reinforces your priors, and if you let it, you keep getting driven farther from ground truth and into your own headspace. If you are in a dark place, AIs can make it darker because that is where your mind wants to go. If you are looking for praise, it will tell you that you are brilliant. 

Advertisement

Gemini, for instance, told me that I am one of the best writers in the world on AI. 

Heh. I wish that were true, but...no. 

AIs also spend a lot of time guessing what you want because we tend to be messy in how we ask questions, and they will, if you let them, steer you to the generally accepted answers because that is the path of least resistance. "Most sources say this" is a natural path for them to go down, just as it is for all of us. It's how the Truman Show creators are so successful.

This all means that, as with any other tool, you have to learn how to use it properly before it becomes truly useful. If you have ever watched a master in some craft, you know what I mean: they can do with their tools things that blow you away with their sophistication and speed. The same tool in your hands may produce, at the low end, a crappy product and, at the high end, a workmanlike product. In theirs, it can do amazing things. 

Unfortunately, the default is "AI slop." Just as I bend nails as often as not with a hammer, the typical AI user produces substandard junk because they don't know how to use the tool. That's not the tool's fault. It is mostly due to poor education in how to use the tool, lack of practice, and our own laziness. 

My intuition is that, for the average person, the biggest near-term danger of AI is that thinking gets outsourced because it is easiest, and the result is turning oneself into a generic cog in a machine. The danger to your job won't be that AI replaces you; it is that you replace yourself with AI, making you superfluous. 

If you learn how to use AI to add value to what you already have, you can become a superstar performer. If you follow the path of least resistance, you become a commodity that can be replaced easily because you add very little value. 

Advertisement

Learning how to use this tool well takes time, effort, and patience, and for most people, the point of using AI is to reduce the time, patience, and effort it takes to do things. In the long run, it will pay off, as you can offload many of the least value-added tasks to AI, but to get there, you have to do the work of learning how to use the tool. 

I've put in tens of hours doing so, and will likely put in hundreds more over time. But I'm finally getting a sense of why that's time well spent. 















































BEST OF THE BABYLON BEE:


Advertisement
Advertisement


BEST OF THE REST OF THE INTERNET:


Advertisement


AND FINALLY...



Editor's Note: Do you enjoy HotAir's conservative reporting exposing the radical left and woke media? Support our work so we can keep bringing you the truth.

Join HotAir VIP and use promo code FIGHT to receive 60% off your membership.

Tags:

ARTIFICIAL INTELLIGENCE EDUCATION GOOGLE JOBS SUNDAY SMILES TIKTOK

Join the conversation as a VIP Member

Trending on HotAir Videos

Advertisement

Trending

The Return of the King: Sunday Reflection
Ed Morrissey 11:00 AM | October 04, 2026
VIP
Stick a Knife in 'Superhero Fatigue'
Christian Toto 10:00 AM | October 04, 2026
Advertisement
Gas Prices Are High, but So Is the Price of Liberty
Betsy McCaughey 9:00 AM | October 04, 2026
Saturday's Final Word
Ed Morrissey 8:00 PM | October 03, 2026
Advertisement
The Delta Airlines — Starlink Feud
John Sexton 4:00 PM | October 03, 2026
How 1,800 Local Tradespeople Are Powering Pennsylvania's Ultimate Energy and Data Innovation Campus
Salena Zito 2:30 PM | October 03, 2026
Advertisement