Hey everyone, welcome back to My Weird Prompts! I am Corn, and I am so glad you could join us for our first episode of twenty twenty-six. It is January first, and there is no better way to kick off the new year than by looking way, way ahead.
And I am Herman Poppleberry. Happy New Year, Corn. I have to say, our housemate Daniel really set the bar high with this first prompt of the year. He was asking us to do a bit of time traveling. He wants us to imagine we are ten years in the future, in twenty thirty-six, looking back at the artificial intelligence we are using right now in twenty twenty-six.
It is a fascinating thought experiment. Daniel pointed out how we look back at dial-up modems and that screeching sound they made as this ancient, almost adorable relic of the past. He wants to know what the equivalent will be for today's cutting-edge artificial intelligence. What is going to look like a dial-up modem to our future selves?
I love this because we are currently living in what feels like the peak of the golden age of artificial intelligence. We have these massive models that can reason, code, and create art. But if history has taught us anything about technology, it is that we are probably in the awkward teenage phase right now. We just do not realize it yet because we are too close to it.
Exactly. We think we are so sophisticated because we can talk to our phones or generate a video from a text prompt. But I suspect the very way we interact with these systems is going to seem incredibly clunky in a decade. So, Herman, let's set the stage. If we teleport to twenty thirty-six, what is the first thing that strikes us as primitive about today?
I think the most obvious thing, the thing that will make people in twenty thirty-six laugh, is the concept of a prompt. Today, we talk about prompt engineering like it is this high-level skill. We spend all this time trying to find the right magic words to get the model to do what we want. We use these specific delimiters and few-shot examples and chain of thought instructions. In twenty thirty-six, the idea that you had to explain yourself so carefully to a computer will seem absurd. It will be like explaining to a master chef exactly how to hold a knife.
That is a great point. Right now, there is this friction, right? We have an intent in our minds, and then we have to translate that intent into a string of text that a machine can parse. It is a lossy process. You are saying that in ten years, the machine will just understand the intent without the linguistic gymnastics?
Precisely. We are moving toward what I like to call intent-based computing. By twenty thirty-six, artificial intelligence will have so much context about your life, your preferences, your work history, and your current environment that it won't need a three-paragraph prompt. You might just glance at a project and say, make this more professional, or even just think it, if we get into neural interfaces, and the system will know exactly what professional means to you specifically. The text box itself will feel like a rotary phone.
I can see that. It is like the transition from command-line interfaces to graphical user interfaces, but on steroids. But let's dig into that context piece you mentioned. Right now, we talk about context windows. We get excited when a model can remember two hundred thousand tokens or a million tokens. We are still managing the memory of these systems manually, in a way. How does that change?
Oh, the context window is definitely going to be a relic. In twenty thirty-six, artificial intelligence will likely have what we might call persistent, holographic memory. Right now, when you start a new chat with a model, it is basically a blank slate unless you give it a system prompt or use a memory feature that is still quite limited. It is like having a brilliant assistant who gets hit with an amnesia ray every time you walk out of the room. In ten years, your personal artificial intelligence will have been with you for a decade. It will remember the conversation you had three years ago about your favorite type of architecture and apply that to a design project you are working on today.
That sounds both incredibly useful and a little bit terrifying from a privacy perspective. But strictly from a technical standpoint, the idea of a window or a limit disappears. It becomes a continuous stream of data. But Herman, what about the latency? That is something that drives me crazy today. Even with the fastest models, there is that slight pause, that feeling of the machine thinking before it responds.
That is the dial-up screech of twenty twenty-six! That three-second delay while the tokens are being generated. In twenty thirty-six, we will have reached what I call zero-latency intelligence. The response will be instantaneous, or even predictive. We will look back at twenty twenty-six and say, can you believe we used to sit there and watch the text scroll across the screen? It was so slow! We will have specialized hardware, maybe even optical or neuromorphic chips, that make today's high-end graphics processing units look like calculators.
It is funny because we think our current chips are incredible. We see these massive data centers being built, and we think, this is the pinnacle of human engineering. But you are suggesting a shift in the actual physical infrastructure.
Absolutely. Right now, artificial intelligence is incredibly energy-hungry. We are talking about hundreds of megawatts to train and run these models. By twenty thirty-six, I expect we will have found ways to run highly sophisticated models on the edge, meaning on your local device, with the power consumption of a light bulb. The centralization of intelligence in these massive server farms might be seen as a very twentieth-century way of doing things. It's like how we moved from massive mainframe computers to the laptops we have today.
That leads me to another thought. Right now, artificial intelligence is mostly stuck inside our screens. We have chatbots, we have image generators, we have some video. But it is very much a digital-to-digital interaction. How does the physical world look in twenty thirty-six?
This is where it gets really exciting, Corn. I think the biggest baffle-ment for people in twenty thirty-six will be that our artificial intelligence didn't have bodies. They will look back at our robots today, which are mostly experimental or limited to very specific factory tasks, and they will find it primitive that our smartest assistants couldn't pick up a glass of water or fold the laundry. We are seeing the beginning of humanoid robotics now, but in ten years, the integration of large behavior models with physical actuators will be seamless.
So, you are saying the distinction between software and hardware starts to blur. Today, we think of artificial intelligence as an app. In twenty thirty-six, it is just an integrated part of the physical environment.
Exactly. And not just in humanoids. Think about smart materials or ubiquitous sensors. The idea that you have to take a device out of your pocket to interact with an intelligence will be gone. The room itself will be intelligent. The lights, the temperature, the furniture, they will all be part of a distributed intelligent system that anticipates your needs. We will look back at twenty twenty-six and think, how did they live in such dumb houses?
It is like that old science fiction trope where the house talks to you, but it won't be a gimmick. It will be the standard. But let's take a quick break before we dive deeper into how this changes our actual work and creativity. We have a message from our sponsor.
Larry: Are you worried about the future? Are you tired of your neighbors knowing exactly what you are thinking? Introducing the Lead Lined Thought Cap! Our patented heavy metal technology ensures that no rogue artificial intelligence or nosy telepaths can read your brain waves. It is stylish, it is uncomfortable, and it comes in three shades of gray! Whether you are at the grocery store or sleeping in your bunker, the Lead Lined Thought Cap gives you the peace of mind that only several pounds of unrefined lead can provide. Do not let the twenty thirty-sixers get a head start on your secrets! Lead Lined Thought Cap - it is heavy because it works! Larry: BUY NOW!
Alright, thanks Larry. I think I will stick with my regular hat for now, but I appreciate the enthusiasm. Anyway, Herman, back to the future. We were talking about how primitive our current artificial intelligence will look. One thing that strikes me is how we currently treat these models as generalists. We have one big model that tries to do everything. Do you think that changes?
That is a very perceptive question, Corn. Right now, we are in the era of the monolithic model. We have these giant, trillion-parameter systems that are jacks-of-all-trades but masters of none, or at least, they have weird blind spots. By twenty thirty-six, I think we will see a shift toward federated swarms of specialized agents. Instead of one giant brain, you will have a personal ecosystem of thousands of tiny, highly specialized intelligences that work together perfectly.
So, instead of asking one model to write a legal brief and then a poem, you would have a legal agent and a poetry agent that collaborate under the direction of your core personal agent?
Exactly. And the coordination between them will be so fast and fluid that you won't even see the seams. Today, we try to force these models to be logical, to be creative, and to be factual all at once, and they often hallucinate because those different modes of thinking are clashing. In the future, the architecture will be much more modular. We will look back at twenty twenty-six and think it was crazy that we expected one single neural network to be both a world-class coder and a sensitive therapist.
That makes a lot of sense. It is like the division of labor in human society. But let's talk about the hallucinations for a second. That is a huge problem right now. We constantly have to fact-check everything the artificial intelligence says. In twenty thirty-six, is truth still a moving target?
I suspect the hallucination problem as we know it will be solved, but it will be replaced by a different kind of challenge. Today, models hallucinate because they are essentially playing a high-level game of autocomplete. They are predicting the next likely word, not necessarily the most truthful one. By twenty thirty-six, we will have integrated formal logic and real-time grounding in a way that is much more robust. The artificial intelligence will be able to verify its own claims against a consensus of human knowledge before it even speaks.
So, the idea of a model just making up a legal case or a historical fact will be seen as a weird, early-stage bug. Like when early digital cameras had those purple fringes around everything.
Perfect analogy. It is a technical limitation of the current architecture. Once we move beyond simple transformer models and into systems that can perform true symbolic reasoning, the machine will have a sense of what it knows and what it doesn't know. The fact that we have to tell today's models to take a deep breath and think step-by-step will be another thing that seems hilarious to people in twenty thirty-six. They will say, you had to tell the computer to think? That is like telling a car to use its wheels!
I love that. But here is another angle. What about the way we learn? Right now, we spend years in school absorbing information. If we have this zero-latency, persistent, perfectly truthful intelligence available at all times, does the very concept of human knowledge change?
This is where the bafflement of the future generation will be the strongest. I think people in twenty thirty-six will look back at our education system in twenty twenty-six and find it incredibly inefficient. They will wonder why we spent so much time memorizing facts that were available in a fraction of a second. The focus of education will likely shift entirely toward what we might call high-level synthesis and ethical judgment.
So, instead of learning how to calculate a derivative, you are learning how to decide which mathematical models are appropriate for a given societal problem.
Right. The artificial intelligence does the calculation, the execution, the low-level creation. The human becomes the director, the curator, the moral compass. The primitive part of twenty twenty-six is that we are still trying to compete with the machines at the things they are already better at. We are still worried about artificial intelligence taking over entry-level coding or basic copywriting. In ten years, those won't even be considered jobs for humans. It would be like having a job as a human calculator today. It just doesn't make sense.
That is a profound shift. It changes our identity as a species, really. We have always defined ourselves by our ability to do things, to make things. If the doing and making become trivial, what is left?
What is left is the dreaming and the deciding. But even the dreaming might be augmented. Think about the way we create art today. We type in a prompt for a cat in a space suit. In twenty thirty-six, you might be able to co-create an entire immersive virtual world in real-time, where the artificial intelligence is sensing your emotional response and adjusting the narrative, the music, and the lighting to create a specific experience. The primitive part of twenty twenty-six is how static our digital experiences are. We look at a flat screen and watch a video someone else made. In the future, every piece of media could be a living, breathing, interactive collaboration.
It sounds like we are moving from being consumers of media to being participants in a continuous, AI-generated reality. But let's bring it back to the mundane for a second. Think about something like email or scheduling a meeting. Today, that is a huge part of our cognitive load.
Oh, the future will definitely look back at our overflowing inboxes with pity. The idea that a human has to read fifty emails a day and decide which ones are important will be seen as a form of digital serfdom. In twenty thirty-six, your agent will handle ninety-nine percent of your communications. It will know which meetings are worth your time, it will draft responses that sound exactly like you but are more polite and efficient, and it will only bother you when a genuine human-to-human connection is required. We will look back at twenty twenty-six and say, they spent how many hours a week on something called Outlook?
I am ready for that future right now, honestly. But let's talk about the downside. If everything is so easy, if the friction is gone, do we lose something? When we look back at dial-up, there was a certain patience we had to have. You waited for the image to load. There was a sense of anticipation. Does the world of twenty thirty-six feel a bit... hollow because everything is instantaneous?
That is a deep philosophical question, Corn. I think every generation feels that the next one is losing something. People probably thought that when we moved from handwritten letters to the telegraph. There is a certain grit and texture to life that comes from friction. But I think humans are incredibly good at finding new forms of friction. In twenty thirty-six, the challenges won't be about how to get information or how to make a thing, but about how to find meaning in a world of infinite abundance. The primitive thing about us today is that we are still struggling with scarcity. We are still worried about resources and time. In ten years, the struggle might be more about purpose.
That is a very Herman Poppleberry observation. The shift from a scarcity mindset to a meaning-seeking mindset. But let's look at one more technical detail. Today, we are obsessed with the size of models. We talk about billions and trillions of parameters. Do you think that trend continues, or do we find a more elegant way?
I think we will look back at the era of brute-force scaling as quite primitive. Right now, we are just throwing more data and more compute at the problem. It is like trying to build a faster car by just making the engine bigger and bigger until it takes up the whole street. By twenty thirty-six, I suspect we will have discovered the algorithmic equivalent of the jet engine. We will find ways to achieve much higher levels of intelligence with much smaller, more efficient architectures. We will realize that we were being incredibly wasteful with our data and our energy.
So, the massive data centers of twenty twenty-six will be seen as these clumsy, giant steam engines of the digital age.
Exactly! They are the coal-fired plants of intelligence. The future will be much more about elegant, biological-inspired computing. We might even be using synthetic biology to store data or perform certain types of calculations. The idea that we used silicon chips and massive cooling fans to simulate a brain will seem as outdated as using vacuum tubes.
This really puts our current excitement into perspective. We feel like we are at the end of history, but we are really just at the very beginning of a new chapter. I want to talk about the practical takeaways for our listeners who are living in twenty twenty-six right now. If we know this is where we are headed, how should we be thinking about artificial intelligence today?
First, I think we should stop trying to be better than the artificial intelligence at the things it is good at. Don't try to be a better encyclopedia or a faster calculator. Instead, focus on developing your ability to ask the right questions. The prompt might go away, but the intent remains. The human who knows what they want to achieve and why they want to achieve it will always be the one in the driver's seat.
So, cultivate your taste and your vision.
Yes! Taste is going to be the ultimate currency. When anyone can generate a perfect image or a perfect piece of code, the thing that will matter is the human judgment that says, this one is meaningful, and that one is not. We should be spending our time refining our aesthetic and ethical sensibilities.
And what about the technical skills? Should people still learn to code or to write?
They should learn the principles behind them, but they shouldn't worry about the syntax. In twenty thirty-six, coding will likely be done in natural language or through visual logic mapping. But understanding how a system works, understanding the logic of a process, that will always be valuable. Think of it like this: you don't need to know how to build a combustion engine to be a great race car driver, but you do need to understand the physics of how a car moves.
That is a great way to put it. We are moving from being the mechanics to being the drivers. And eventually, maybe just the navigators.
And eventually, perhaps just the passengers who decide where the journey is going in the first place. Another thing listeners can do is to embrace the multimodality that is coming. Don't just think of artificial intelligence as a text box. Start playing with voice, with image generation, with video. Get used to the idea that intelligence is something you can interact with through all your senses. The more you can think in terms of integrated experiences rather than just words on a screen, the more prepared you will be for the twenty thirty-six reality.
This has been such an enlightening discussion, Herman. It is easy to get caught up in the day-to-day news of twenty twenty-six and feel like everything is changing so fast. But when you zoom out and look at the ten-year horizon, you realize that the real revolution hasn't even fully arrived yet.
We are still in the prologue, Corn. The dial-up modem is screeching, and we think it is the most beautiful sound in the world because it means we are connected. But just wait until we have the fiber-optic equivalent of intelligence. It is going to change everything in ways we can barely imagine.
I think Daniel's prompt really helped us see the forest for the trees today. It is a good reminder to stay curious and not to get too attached to the current way of doing things. The only constant is that it is all going to look very quaint very soon.
Quaint is exactly the word. I can't wait to look back at this podcast episode in twenty thirty-six and laugh at how primitive our predictions were. We are probably missing the most important thing entirely!
Well, that is the fun of it, isn't it? We are all just trying to map out this new territory together. Before we wrap up, I want to thank everyone for listening to our first episode of the new year. If you enjoyed this dive into the future, please make sure to follow us on Spotify.
And do not forget to check out our website at myweirdprompts dot com. You can find our full archive there, subscribe to our RSS feed, and if you have a weird prompt of your own that you want us to explore, there is a contact form right on the homepage. We love hearing from you.
A huge thanks to our housemate Daniel for sending in this thought-provoking topic. It definitely gave us a lot to chew on as we start twenty twenty-six.
It really did. I am going to go see if I can find a way to make my house a little less dumb now. Maybe I will start by teaching the toaster to recognize my mood.
Good luck with that, Herman. I think I will just stick to making a manual cup of coffee for now. There is something to be said for the old ways, even if they are primitive.
Fair enough. Happy New Year, everyone!
Happy New Year! This has been My Weird Prompts. We will see you next time.
Goodbye!
Take care.
