Creating human-level intelligence is not just allowed by the laws of nature, we even have an existence proof: people.

While we are only at the beginning of incorporating AI into scientific processes, AI has already found adoption in a wide range of scientific domains where it is accelerating the discovery of new explanatory knowledge. The next phase in the evolution of the scientific process will likely see (semi)autonomous AI scientists that radically transform the pace at which science is done.

Some AI safety challenges are not technical problems. They might not have a technical solution, but instead they pose novel moral dilemmas simply because AI is taking actions in the world.