The Digital Frontiers Showcase celebrated the fact that 2023 marked the sixtieth anniversary of Computer Science and Artificial Intelligence activities at the University of Edinburgh.
The showcase consisted of talks and panel discussions with the University’s world leading experts in data science and digital technologies including AI, Quantum Computing, robotics and novel chip architecture.
The programme provided a realistic roadmap for some of the profound long-term impacts that the research that is pushing the boundaries may bring for businesses and society. It also covered some of the novel solutions and capabilities that will be deployed in the near term to transform our approaches to the climate crisis and fundamental challenges in health and healthcare.
In this live essay I provide my current thinking on the central question: “Which new technologies are pushing the boundaries?”
Natural Language Processing
We have crossed a threshold where “Natural Language Processing” is no longer an accurate descriptor; we are now in the era of Natural Language Action. As Mirella noted in the panel, the true revolution isn’t just that these models can write poetry or code—it’s that natural language is becoming the universal interface for computing.
The shift from understanding text to generating executable logic is profound. We are seeing models that don’t just output text but function as “controllers”—orchestrating other tools, APIs, and systems. The “neuro-symbolic” dream of the past is being realised not by hard-coding logic rules, but by teaching LLMs to call upon specialised modules (calculators, search engines, simulators) when their probabilistic intuition isn’t enough.
My Prediction: The “One Model to Rule Them All” era is already ending. The sustainability constraints highlighted in our discussion—the massive energy cost of training and fine-tuning giant models—will force a fragmentation. We will move towards a “modular mind” architecture. Instead of training a trillion-parameter model to know everything, we will see smaller, hyper-specialised “reasoning engines” that act as conductors, coordinating a symphony of expert tools. The future isn’t a smarter chatbot; it’s an operating system where you speak your intent, and the OS negotiates with a fleet of specialized agents to execute it. I’m already seeing this with my stack of tools including WisprFlow and Cursor.
Quantum Computing
Quantum is not just “faster computing”—it is a fundamental rewriting of the laws of information. As the panel discussed, treating Quantum as a mere hardware upgrade is a category error. It is a new way of thinking that requires us to abandon classical intuitions about certainty and state.
The “Dream and the Threat” duality is real and pressing. The “Threat” (RSA encryption collapse) often dominates the headlines, but the “Dream”—quantum simulation—is where the revolution is already quietly happening. Industries from Rolls-Royce to chemical giants are effectively using quantum (or quantum-inspired) simulations to model physical reality with a fidelity classical systems cannot touch. The immediate value isn’t in cracking codes, but in cracking nature itself: designing new catalysts, batteries, and materials by simulating them at the sub-atomic level.
My Prediction: Quantum Privacy will arrive before Quantum Supremacy. While we wait for the fault-tolerant quantum computer that can break the internet, we will see the rise of “Physics-Based Trust.” As highlighted in our panel discussion, quantum technologies offer security guaranteed not by mathematical complexity (which can be solved) but by the laws of physics (which cannot be cheated). We will move from a “trust-based” internet to a “physics-based” internet, where data sovereignty and privacy are enforced by the very nature of the particles carrying the information.
Robotics
We need to kill the “Humanoid Fallacy.” The public obsession with robots that look like us is, as I bluntly put it, a “lazy design heuristic” that holds the field back. The future of robotics is not metal men; it is Physical Intelligence.
The most successful robots today—from Amazon’s fulfillment swarms to soft-robotic grippers packing fruit—succeed precisely because they don’t try to mimic humans. They embrace “morphological computation,” where the intelligence is embedded in the body itself. A soft, octopus-inspired tentacle doesn’t need complex force-feedback sensors to pick up a strawberry; its material properties solve the control problem for free. This is the shift from “software controlling hardware” to “hardware that thinks.”
My Prediction: The “Invisible Robot” Revolution. While the media focuses on bipedal robots doing backflips, the real revolution will be pervasive and invisible. We will see the rise of “infrastructure robotics”—self-repairing pipes, swarm-based agriculture, and soft-robotic medical tools that navigate our veins like catheters. These machines will look less like C-3PO and more like slime molds, tentacles, and swarms. They will not be “assistants” we talk to; they will be the environment itself, a living infrastructure that adapts, repairs, and responds to our needs without us ever making eye contact.