Can AI Make a Cup of Coffee?
Artificial General Intelligence, or AGI, is the hypothetical point at which AI can match or exceed human abilities across virtually any task—not just answer questions, write code, or generate images. Unlike today’s narrow AI systems, which excel within defined domains, an AGI would need to understand new situations, transfer knowledge between contexts, and act effectively in the physical world.
Why coffee matters
One of the most memorable ways to explain that gap is the “coffee test,” associated with Apple co-founder Steve Wozniak. The challenge sounds almost trivial: send an AI-powered robot into an ordinary home it has never seen and ask it to make a cup of coffee.
But the task is deceptively difficult. The machine must navigate unfamiliar rooms, recognize a kitchen, locate a coffee maker, find coffee and a mug, understand cabinets and drawers, add water or grounds where needed, press the right controls, and avoid spilling or breaking things. It requires perception, mobility, planning, common sense, and real-time problem-solving—abilities that remain hard to combine in one system.
AI is impressive, not human
Today’s AI can draft a business plan, analyze sales data, translate languages, and even explain how to brew espresso. Yet these systems mostly operate through text, images, or carefully controlled environments. They do not naturally possess the embodied understanding a person uses when entering a strange kitchen and improvising.
That is why the coffee test is useful for business readers: it separates genuine general intelligence from impressive automation. A chatbot may know the recipe for coffee, but knowing is not the same as finding the mug, interpreting a messy countertop, and completing the task safely without prior instructions.
The business takeaway
For companies, AGI is not simply “better AI.” It would mean systems capable of handling unpredictable, end-to-end work in the real world—customer service, logistics, manufacturing, or office operations—without narrow reprogramming.
Until AI can reliably walk into an unfamiliar home and make a cup of coffee, claims that it has become human-like should be treated with caution. AI is transforming business, but the coffee test reminds us that human-level general intelligence remains a goal, not a finished product.