Post #1902114

@festal I look forward to when people's idea of AI is no longer throwing stupid amounts of LLM compute at a problem, but actually using the many, many other task-specific AI models, which are usually also far more efficient