Apple made headlines recently, but not in the way most people expected. The tech giant released a whitepaper that claimed AI does not think. The internet erupted with surprise, but for those working closely with AI, this was not a revelation. It was something we already knew.
In episode 009 of the Dootrix podcast, Rob and Kevin dissect what Apple actually said, what they left out, and what it all might mean for the future of software, hardware, and AI adoption.
The central argument in Apple’s whitepaper, The Illusion of Thinking, is that large language models are prediction engines. They are excellent at low and medium complexity tasks but falter when faced with high complexity challenges. This mirrors what many AI researchers have said for years. LLMs are not reasoning machines. They generate plausible outputs by spotting statistical patterns.
So why the fuss? Partly because Apple’s timing felt strategic. While others have been building massive cloud-hosted models, Apple has lagged. Now they are positioning themselves as the champion of a different path: small, secure, on-device models that prioritise privacy. The paper reads less like a breakthrough and more like a justification.
Although the paper raised some fair points, it missed a crucial trend. The wider AI community is shifting towards agentic systems. Rather than asking a single model to solve a complex task all at once, developers now split that task into smaller chunks and assign each to a specialist agent.
Kevin highlights that if LLMs perform well at medium complexity, this suits agentic approaches perfectly. These systems stitch together many smaller interactions, making the most of what LLMs do well. Apple’s paper fails to acknowledge this direction. It critiques a problem that others have already begun to solve.
WWDC also brought news of iPadOS 26. On the surface, it looks like progress. It finally brings the iPad closer to being a full computer. But Rob questions whether that is a good thing.
The original iPad was a vision of the future. It was simple, focused, and designed to do one thing well. The dream was to remove clutter, eliminate complexity, and make powerful computing accessible to anyone. The new update leans back towards a traditional desktop experience. It risks turning the iPad into just another touchscreen Mac.
Kevin argues that this change may reflect a more tech-savvy public. Expectations have shifted. Users want flexibility and power, not just simplicity. Still, the shift feels like a missed opportunity. Rather than using the iPad as a platform to reimagine what computing could be, Apple has chosen to merge it with the past.
Apple’s big design announcement was the new Liquid Glass aesthetic. It looks sleek. It blends well across devices. But it is hardly groundbreaking. Some have pointed out it resembles the Aero Glass interface from Windows Vista.
Others suggest Apple is playing a longer game. Liquid Glass may be a small step toward normalising layered, semi-transparent interfaces. These could serve as a bridge between today’s flat screens and tomorrow’s mixed reality environments. Still, for many, the reveal felt underwhelming. In a world moving toward screenless interactions and voice-first computing, Apple’s biggest announcement was still about the screen.
The conversation then turns to Mistral, a French AI company backed by the EU. Their latest model, Magistral, is designed with compliance in mind. It is multilingual by default and aims to meet the strict demands of sectors like finance, healthcare, and government.
Unlike many flashy models, Magistral is focused on auditability. It allows full traceability of how the model arrives at its conclusions. In a world that increasingly needs explainability and control, this model could carve out a valuable niche.
Towards the end of the episode, Rob raises a provocative idea. Could Tesla be the biggest AI company on the planet? While OpenAI and others focus on text generation, Tesla is deploying AI in the physical world. Their robotaxi programme in Austin is moving into public trials. Soon, Tesla owners may be able to add their vehicles to a shared fleet, allowing others to summon and use them autonomously.
Kevin notes that Tesla has always been more than a car company. Its real asset is the AI that helps vehicles understand and navigate the real world. That same AI could power humanoid robots or future space missions. With real-world deployment and potential for massive consumer revenue, Tesla’s valuation may still be on the low side.
Despite the hype, what counts is practical application. AI is already changing how software is built, how teams work, and how businesses operate. Whether it is called a chatbot, an agent, or a robot, the focus should be on outcomes.
Rob and Kevin agree that the promise of AI is not just to make work easier. It is to remove the work altogether. That future may still be forming, but its foundation is already in place.
Even if we are still waiting for a robot that can put the washing away.