This article provides good examples of how the efficacy and efficiency of a given technology is often less important than deeper questions of reliance and roles. link to ‘Too much trust in machine translation could have deadly consequences.’

Gotta admit that I’d never thought about what we should do about algorithms trained on data that’s subject to a deletion request. Interesting article. link to ‘Now that machines can learn, can they unlearn? | Ars Technica’