An AI model has achieved human-level performance on general intelligence test --what does this mean for the future?