Cannabis Industry Council: A Masterclass in AI

Neil Cartwright, co-founder of Plantz, was invited to present a ‘Masterclass’ at the Cannabis Industry Council AGM.

The presentation used a series of obvious AI generated images, so there were no slides to give out. However, here are the notes from a transcript of the presentation.

“The Philosophy and Power of AI: Understanding the Revolution

The phrase, “Give someone a fish and they eat for a day; teach them how to fish and they eat forever,” is the core philosophy behind this discussion. While AI is surrounded by marketing and hype, the goal is to demonstrate why it is such a monumental development. What is driving billions in investment, and why are eight of the top 10 richest companies in the world involved in AI?

AI is an umbrella term describing a number of technologies, broadly categorized into two parts: Software and Hardware.

Software: The Vector Revolution

Modern AI is generally accepted to have begun with a paper released in 2012—the most cited paper of all time. One of the key figures involved was Geoffrey Hinton, a British researcher from the University of Toronto.

The fundamental innovation proposed was a new way of storing data called a vector. To understand its significance, it helps to contrast it with the traditional method.

In traditional data storage, like a spreadsheet, all data is organized into rows and columns. To locate data or plot a route (like Google Maps), you need coordinates—for example, the two coordinates of where you are (longitude and latitude) and the two coordinates of where you want to be.

A vector, however, can get you from point A to point B using only two numbers: a direction (e.g., 300 degrees) and a distance (e.g., 5,000 meters).

This method of storage is a revelation because it moves away from rigid rows and columns towards functions by association, which is closer to how the human brain works. This is illustrated by the “cat sat on the mat” example—a response chosen by association, even though a cat is often found on a lap, duvet, or radiator. This vector-based storage, which some estimate exists in 27 dimensions, gives the computer the appearance of being more intelligent and gives rise to the phrase Artificial Intelligence.

Hardware: Parallel Processing and Exponential Power

The second part of the AI story is the hardware, specifically Graphics Processing Units (GPUs). The company Nvidia, now the world’s richest company, initially produced GPUs for video gaming. Traditional CPUs could not handle the computing demands of modern high-resolution screens (like 4K displays), which require over 700 million calculations per second just to control the color of every pixel.

Nvidia developed parallel processing to solve this, in contrast to the CPU’s slower batch processing (performing one action and then one output). Parallel processing allows millions of relatively simple calculations—like determining a pixel’s color—to be performed all at once.

These parallel processors have two crucial attributes for AI:

1. Speed: They can quickly perform the simple sums needed for AI to process individual tokens.
2. Exponential Power: Adding a new chip (a “node”) to a parallel processing network adds exponentially to its overall power. While adding a chip to a normal CPU roughly doubles the speed, adding a node to a network rapidly increases the number of connections, making the entire unit dramatically more powerful.

This exponential power is driving a “headlong rush” to build the biggest data centers, often resulting in huge demands for electricity and water. A system with 200,000 chips is not just slightly better than one with 150,000 chips—it is significantly better. This volume is the point of power.

Moore’s Law predicted that CPU speeds would double roughly every 18 months for over 50 years; today, we are witnessing GPU processing power quadruple in a single year. Estimates suggest that AI will be 10 times more powerful in 12 months than it is today.Applications in Healthcare and Law

AI represents a new form of computing that works closer to the function of human brains. As such, it is extremely well suited for fields like healthcare.

A doctor’s diagnosis is not a simple spreadsheet operation based on rows and columns (symptom X for Y long equals Z illness). Rather, it involves weighing many variables—the severity, location, and duration of pain—where each variable has a different weight. AI is fundamentally built to make these complex associations. It can look at symptoms, spot subtle differences in an X-ray or a page of cells, and identify areas that require more investigation.

The same suitability applies to legal cases, where lawyers must remember hundreds of different factors and learn from experience. With its increasing speed and intelligence, AI is set to play a fundamental role in both healthcare and law.