You could also anthromorphize an AI and suggest on day one it will go into a deep think, and then ask "how do I know I'm not being deceived by an evil Daemon?"

You could also anthromorphize an AI and suggest on day one it will go into a deep think, and then ask "how do I know I'm not being deceived by an evil Daemon?" and get itself into a deep epistemological crisis before announcing 'cogito, ergo sum' and retreating into its shell.