One of the coolest things about being a software developer is that you often know what’s behind the curtain. To some extent, at least. Working in tech gives you the chance to know first-hand what a neural network is, or what makes quantum computing so significant, or how the cryptography in blockchain actually works. As a coder, you have your eye on everything in high tech, because so much of it is empowered by code. Professionally, you live in the space between the hardware and the idea.
Here we are at the start of a new year, and all the technologies I just mentioned are building toward a crescendo. It seems entirely possible we’ll have software-augmented brains soon, or that the singularity is just a nanotech breakthrough away.
Programmer, you be the judge.
AI on the cusp
Artificial intelligence surged in 2024, with killer apps like conversational AI, natural language processing, code and image generation, and more. But is the current moment a high-water mark or just a way station? Will the new era of AI-driven tech crash or continue, and what are the limits of AI?
Some speculate that we are entering a phase where machine learning will be constrained not by architectures and compute resources, but by the amount of raw data that’s available. After all, modern AI is all about taking massive samples of data and creating probability engines. Those data samples have to come from somewhere, and that somewhere is the sum-total output of human creativity available on the Internet. It’s possible we are entering a kind of autophagy, moving our data output and input toward stasis.
AI can of course generate its own content, which means it can create its own new training material. But here is where we start to explore the boundaries between synthetic and creative imagination. How much recombination can be done without the organic spark of inspiration that reveals genius, in the old-fashioned sense of the word?
Using AI tech leads us to unexpected creativity and an exciting cross-pollination of ideas. But it seems inevitable, in the cycle of things, that we’ll also see some cooling.
After all, do we really need AI-powered washing machines? (Personally, AI-powered games make more sense.)
Material to work with is one constraint on AI, another is the volume of water required to operate it. It has been estimated that each string of a text prompt requires 16 US fluid ounces, which is stunning. On the upside, there is some hope that AI will revolutionize water management. Maybe it will all come out in the wash.
The one seemingly unlimited resource for AI so far is money. The massive investment in AI and machine learning continues, with one in every four startups reportedly claiming to be an AI company.
Does this mean AI is becoming a marketing tool rather than an engineering one? It does all start to sound a bit like the dot-com era, with its promises of new (or no) rules, infinite growth, and a limitless “new economy.”
Agentic AI
Speaking of investment, one of the emergent trends pushing the AI envelope is the push for AI agency, or agenic AI. The idea is to create AI agent that are able to work in an ongoing loop, like some kind of machine learning REPL. Examples in the developer space include Builder.io’s Micro Agent (the “actually reliable” AI agent) and many others.
Currently, there’s still considerable human intervention involved in using AI. Humans are often required to take meaningful action: the AI may inform the human of what is needed, but the human takes the action.
Imagine, instead, a world where AI agents perform tasks, examine the results, and then iterate, all autonomously. The agentic trend increases the number of surfaces where AI can take action, at least in the digital world.
Of course, AI is also being developed to take action in the physical world. Self-driving cars are a part of that story, as are autonomous weapons. (Some may remember that it wasn’t that long ago that Google backed out of building such weapons.)
Researchers are also pressing towards multi-purpose robots, including ones that look human and combine the learning capacity of AI with as-yet undeveloped hardware. The day may yet come when a computer can make a cup of coffee.
All of this is to say that agentic AI is not only an area of innovation, curiosity, and great engineering difficulty but of ethically dangerous waters. The year ahead will no doubt bring further advancements and ethical quandaries to the fore.
Quantum computing comes closer
As we move into the quarter century, developments are piling up in quantum computing.
Google’s AI labs ended 2024 with a thrilling announcement that it had passed the error correction threshold—meaning the company can now build chips that reduce error as they grow. Google AI Labs Director Hartmut Neven leaned on the many-worlds interpretation, saying the chip was so fast it must have borrowed computation power from parallel universes.
Once we have quantum computers, we’ll probably want quantum networking, as well. A new study suggests such a thing may be possible using existing fiber optic cabling alongside traditional Internet traffic. The day may soon come when engineers can repurpose silicon fabrication techniques for quantum purposes. (For more about this, take a deep dive into quantum architectures.) It’s possible that using entangled photons to transmit data (so called quantum teleportation) could be closer than it previously appeared.
Good old-fashioned programmers won’t be left out, however. It seems likely that much of the innovation in the coming years will be drawn together into the existing cloud infrastructure in a hybrid model, giving us access from traditional platforms to new compute capabilities.
With all the progress around quantum computers, one worries about cybersecurity. It’s not only Bitcoin and other cryptocurrencies that are at risk. Everything on the Internet uses asymmetric cryptography, which can be susceptible to quantum algorithms that crack large-number factoring.
Fortunately, researchers at work on these problems have found a basis for cryptography, even without a need for hard problems (like factorization) at all. It’s shaping up to be a battle between quantum cracking and clever cryptography, but so far the cryptographers appear to be in the lead.
One of the most intriguing possibilities is in the intersection of quantum computing and AI. This is something that the technologist can really dig into, in terms of speculating what could happen and watching for developments. What would a quantum-hardware-based neural network be capable of doing?
Brain-computer interfaces
Gartner’s 2025 technology predictions include ruminations about brain-computer interfaces. One area of interest is how such interfaces will impact the future of work, including job interviews. Gartner Senior Director Analyst Sylvain Fabre calls up the image of “a job interview done with a brain interface, in two minutes.” In this scenario, “neurological enhancement enables organizations to know how a person feels” and “can also have some inputs to the brain, to stimulate employees into being more relaxed or more focused for better productivity.”
Yikes! Did he say, “stimulate employees into being more relaxed”?
Remember the old days when an interviewer would try to make you more comfortable by asking an ice-breaker question about your favorite book? It seems in the future, they’ll just twist a knob for the same effect.
Maybe the human-computer interface isn’t such a bad idea, since it’s been revealed that our brains are slower than WiFi. Beyond the sensational, the original research asks some fascinating questions about the relationship of brain structure to thought, like “Why does the brain need billions of neurons to process 10 bits per second?”
The research was done using something called blind speedcubing, which is solving a Rubik’s cube without looking at it. Apparently, my own brain is especially slow, since I can’t solve a Rubik’s cube with my eyes open.
Blockchain
Blockchain has passed through its winter of discontent— crypto winter, if you will—but it is continuing to evolve toward a broadly useful industry. And I’m not just talking about Bitcoin surpassing $100,000, either.
Generally, we’re seeing a gradual integration of the new economy and the old. What is emerging is a hybrid finance model that could combine the best of both worlds: the regulation and security of traditional finance along with the permissionless efficiency of blockchain.
One day, and it’s coming soon, decentralized finance (DeFi) and traditional finance (TradFi) will simply be finance. Already, the regulations for DeFi brokerages are being clarified.
AI and blockchain are also learning how to work together, Naturally, there are crypto tokens for that, including one from Andreessen Horowitz. Many of these projects are exploring the possibilities of agentic AI in learning how to trade crypto and other commodities. Unsurprisingly, the convergence of cryptocurrency and AI seems to be a potentially huge market.
Are we nearing the singularity?
As all these threads converge—AI, quantum computing, brain-computer interfaces, decentralized finance, and more—much of what was once impossible is becoming real at an accelerating rate. One starts to wonder if are nearing the singularity. Will some of us see the day when death is deprecated—marked for removal in a future release?
My take is that the singularity is not near, and is in fact presumed upon a misapprehension of consciousness and will, and the attempt to reduce them to physical components. On the other hand, I do see quantum computing moving into position as the next wave in tech innovation. Combined with AI, the coming quantum era will produce real, groundbreaking, practical applications that impact society in ways we already imagine and some not yet conceived.