Environmental (Commonwealth Union)_ In a dimly lit Sydney office with a tang of old books and rich coffee, Australia‘s beloved science communicator is preparing something new a digital clone of himself that’s purpose in life is to battle climate denial. Dr. Karl Kruszelnicki, the 77-year-old physicist-turned-national-icon known for his madly striped shirts and boundless enthusiasm for facts, is creating an AI clone of himself to cope with the 300 daily climate questions he can never seem to tackle on his own.
The project, called Digital Dr Karl, is both a high-risk gamble and a technological marvel a gambit to use artificial intelligence against the blizzard of climate misinformation Kruszelnicki asserts has been “pumped into the public like a firehose for 30 years.” Powered by an open-source language model that has learned from 40,000 PDFs in the scientist’s own library, the chatbot doesn’t just regurgitate facts it mimics Dr. Karl’s trademark humor, warmth, and annihilating reason.
“Imagine having an argument with someone who’s learned all the climate studies ever published and doesn’t fatigue,” explains Leigh Stark, the tech journalist collaborating on the project. During a demo, the AI still in gangly teens mode is right to identify fallacies in the “climate hoax” myth but sometimes stumbles, confusing metaphors or inventing data points. But initial research suggests that even faulty AI can alter minds; a recent Science study found chatbots reduced conspiracy theory following by 20% with impacts lasting for months.
The stakes are as high as they can be. With climate disinformation spreading at record speeds, Kruszelnicki’s crew is working against the clock to build something that’s essentially a fact-based immune system a digital white blood cell against viral lies. The AI exists on a solar-powered Mac in an attempt to balance its carbon footprint, a reference to the irony of combating an energy crisis with energy-hungry technology.
By October, Digital Dr Karl will be subjected to its ultimate test: 100 days of public interrogation by skeptics and curious minds alike. The team will be monitoring each exchange, ready to “pull the plug” if the bot strays from message. For Kruszelnicki, it’s not an experiment rather, payback for 16 years’ worth of free schooling. “If I can change even a few minds,” he says, pushing his trademark glasses up his nose, “that’s more than another book royalty check.”
In a time of AI risk, drowning us in its simulated messaging, there is one scientist who is taking a risk that it can also throw us a lifeline if only it’s given enough evidence, ethics, and goofy Hawaiian shirts. It’s not if machines can think, but if they can get people to think again.
