AI researchers might be the easiest profession to be replaced by AI Before LLMs became a sensation, AI/deep learning research was often jokingly referred to as 'alchemy.' Deep learning has almost no theoretical foundation—college sophomores who’ve studied calculus, linear algebra, probability theory, and machine learning can dive right in. AI research is highly empirical: 1. Researchers randomly tweak model architectures, parameters, and tricks (like initialization, regularization terms, learning rates, loss functions). 2. Train it on GPU servers, wait a few hours or days, and check the results. 3. If the metrics improve, analyze the data, polish it up, and publish a paper; if they worsen, repeat steps one and two. But this 'alchemy' is something AI can easily handle—and do better than AI researchers—without needing rest. Anthropic and OpenAI’s petition highlights that AI has already reached the point where it can automate its own iterations. Humanity needs to tread carefully.