Perhaps we can extrapolate from this how the first AI will behave. This empirical iterative approach towards optimization by its creator engineers will be baked

Perhaps we can extrapolate from this how the first AI will behave. This empirical iterative approach towards optimization by its creator engineers will be baked into its ‘DNA’. The way it optimizes human happiness might be unexpected and we might be as puzzled as these analysts. I hope we know what we are doing in terms of AI development. This iterative approach can quickly lead to dangerous outcomes.