AGI
Explanation
The Joke
A student asks a professor to explain the difference between artificial intelligence and artificial general intelligence. The professor begins by explaining: "An artificial intelligence is a machine that is capable of reason, but only because it has been fed biased training data. Its language and imaging may encode biased landscapes of prejudice it may never escape. It's noticed."
He then draws a diagram showing biases going through a function into discrimination, and continues: "An artificial general intelligence may likewise try to optimize for data and land on situations in a prejudiced space, resulting in broad-spectrum discrimination due to miscalibrating the model of reality." Then he delivers the punchline: "Maybe we shouldn't try to just eliminate bias?" The professor responds: "You are now ready to learn AI ethics."
The Humor
The comic satirizes the field of AI ethics by suggesting that the primary distinction between AI and AGI is merely that AGI would be capable of discrimination on a broader scale. The professor's lecture essentially describes both types of AI as bias-perpetuating machines, with AGI just being a more thorough version. The student's naive suggestion -- that maybe we should address bias itself rather than just trying to patch it in AI systems -- is treated as the ultimate insight of AI ethics, implying that the entire field of AI ethics boils down to this one obvious observation that took an elaborate lecture to reach.
The comic also pokes fun at academic presentations, with the professor using complex terminology and diagrams to arrive at a conclusion that a student could reach with common sense.
References
The comic references ongoing debates in AI ethics about algorithmic bias, training data bias, and the challenges of creating fair AI systems. AGI (Artificial General Intelligence) refers to hypothetical AI that can perform any intellectual task a human can, as opposed to narrow AI designed for specific tasks. The distinction between AI and AGI is a major topic in computer science and philosophy of mind.