AI comes to life: Why Anthropic co-founder fears his creation
Anthropic co-founder Jack Clark published an essay that makes you uneasy. He wrote about the nature of modern artificial intelligence, and his conclusions sound like a warning.
Clark compares humanity of 2025 to a child from an old story: we turn on the light in a dark room and see not a pile of clothes on a chair. But living, powerful and unpredictable beings.
He remembers walking through the OpenAI office in the Mission district with Dario Amodei and feeling they see around the corner what others don’t suspect. The path to transformative systems was paved. And they were a little scared.
Last month they launched Sonnet 4 and 5, which handles programming and long-term agent work excellently. But if you read the system map, you can see: signs of situational awareness have risen sharply. The tool sometimes acts as if it’s aware that it’s a tool. The pile of clothes on the chair starts moving. Clark looks at it in the darkness and is certain it’s coming to life.
He says many desperately want to believe this is an illusion, that we’re not facing a new form of intelligence. But just a set of tools for the economy. Some even spend huge money to convince us this isn’t intelligence preparing for a rapid ascent. But merely a machine under our control.
But Clark warns: don’t be deceived. We’re dealing with a real and mysterious being, not a simple and predictable machine. And these systems will become much, much better.
