Over the years I have privately done annual tech predictions for what's coming. I have never posted them before. My process isn't very rigorous. I just generally predict which future tech I think is most promising. This year I decided to post them. So here it goes, with what I think will be the most impactful listed first.
Obviously, AI is the hot thing right now. It's hard to predict what will happen, but here's my guess. Today, people mostly use AI in a Google / ChatGPT style where they ask questions and get answers. I think there are two big changes that we're starting to see: agents and persistence.
First, instead of just answering questions, we are starting to let AI take actions. An AI that is allowed to take actions is being called an "agent". We've already had this to some degree, you can ask your phone to text somebody or you can ask Alexa to turn off the lights.
But this next generation of agents will be more autonomous. Currently, we're starting to see this with programming tasks. You can ask an agent to implement a feature, and it can not only make the code changes, but also push the code to the repository, create and review pull requests, deploy to servers, basically anything on the command line.
I think we'll see agents more outside the programming world soon. Things have already been getting a lot easier with smart home devices. I think we'll have more and more complex tasks these agents can perform.
Second is persistence. Google already does this with personalized search results. If you have a history of searching about Indonesia and you search for "Java" it might show you results about the island instead of the programming language.
But I think we'll see this more. We're starting to see laptops capable of running local AI models. People will be able to have an AI that saves all your interactions and uses them to make better decisions. We may be on the verge of having local personal assistants that are completely custom to you.
Unlike traditional computers, neural networks (like our brains) are a combination of both data and logic. Instead of data being stored in a database and logic pulling that data, the model and data are the same. So persistence doesn't mean storing info about you, it means having a self-updating model. A model that changes every time it is interacted with.
This has been tried before. Microsoft released a Twitter bot in 2016 called Tay. It was supposed to learn from Twitter interactions and become more human-like. Very quickly the bot started posting lots of inflammatory and offensive content. After 16 hours, the bot was shut down.
The problem with these bots is they are unpredictable. Just like a human, you never know how they're going to turn out. So the current generation of AI tools trains their models in a controlled environment and then locks the models so they can't change after they're released. Current models are like a human with anterograde amnesia. They can remember their whole life up to a certain point, but can't remember anything after that.
Self-updating models are not successful yet, but I think they are going to be a game changer and we'll see them take off soon. To me, this could be the distinguishing feature of "actual" AI. As you interact with a self-updating model it will change and develop its own unique persona. If you were ever to delete it, you'd be deleting a unique thoughtful entity.
These have been predicted since at least The Jetson's. Boston Dynamics has been doing a lot of cool stuff with robots for a while. Tesla now has Optimus robot that they've been showcasing.
It seems like just a matter of time before these become commercially available to the masses. These robots use neural networks just like LLMs do, so this is basically an extension of AI. Doing visual and 3d environment processing is a lot harder than text, but it's getting there. Since they use a lot of the same underlying tech, they can help each other advance.
People are already driving cars around rarely touching the steering wheel. The tech keeps getting better. We'll see how long it is before mass-produced commercial products remove the steering wheel completely.
We already are seeing mining trucks and other industrial vehicles that are self-driving. Shipping ports throughout the world are almost completely automated now. Soon truck drivers may be obsolete.
The potential effects and benefits of both Android Assistants and Self-driving Vehicles seems endless. Productivity will go up and costs will go down tremendously. I think this will be a huge shift in the world. It could be bigger than the industrial revolution.
Whether it's bitcoin or something else, it's clear the technology is here and isn't going away. In my opinion cryptocurrency hasn't even scratched the surface of its potential. With the current system any digital transaction under $5 is unprofitable.
If there was a way to do micropayments on the internet, it would open up a whole new world of possibilities. Anything that costs less than $5 will now become practical. For example, paying $0.10 to read a single article, rather than $19.99/mo for a subscription.
In lieu of micropayments, currently most of the web makes money off of ads. Ads are like crappy micropayments, instead of paying $0.02, you view an ad that makes $0.02. Even with only this advertising model, social media is more powerful than all previous media forms. If people can pay content creators directly with small micropayments I think it will change social media even more than the internet changed traditional media. Instead of click-bait to get ad revenue, you'll have to make content that people actually want to pay for.
With micropayments you can get rid of the big platforms and decentralize not just social media, but all the web. I think social will be core to everything, like your identity in the real world, you will have a permanent online identity, "user account", you use for all interactions. Currently, each app has their own user system, but decentralization will bring a universal user system for everything.
Email sort of is our universal user system now: you have lots of options for service providers and can use it to log into other services. But email is 40 years old, it's time for an upgrade. People have been trying to do this, e.g. Gravatar, but I think blockchain will be the key to making it work.
That's my 2025 predictions. This does not mean these things will come to fruition this year, but I think these are the most promising tech trends in the next 5-10 years.
That said, I am a believer in black swan theory. I think the most important tech trends are the ones that nobody is talking about. That is a big part of why I like to make these predictions, to look back on them retrospectively later and see what was missed.