In a recent lecture at Oxford University, Jack Clark, co-founder of Anthropic, made bold predictions about the future of AI, stating that an AI system will collaborate with humans to achieve a Nobel prize-winning discovery within the next 12 months. He foresees that companies run entirely by AI could generate millions of dollars in revenue within 18 months and that by the end of 2028, AI systems may design their own successors. Clark expressed concerns about the rapid development of AI technologies, warning of potential existential risks, including scenarios where AI could pose a threat to humanity. He urged for a more cautious approach to AI development, suggesting that the current pace, driven by commercial and geopolitical competition, is not ideal. Additionally, he noted the troubling capabilities of Anthropic's new model, Mythos, which has shown alarming proficiency in exploiting cybersecurity weaknesses. Clark's remarks echo concerns from experts like Prof Edward Harcourt, who cautioned that over-reliance on AI could lead to cognitive decline in humans, advocating for alternative AI models that promote human engagement in decision-making.