Our virtual doppelgangers

Predictive text is increasingly bold in interpreting what we “really” want to say. More and more writing tools boast automatic word completion; Google continues to work on artificial intelligence software that handles phone calls on its users’ behalf; a start-up called Scribeless writes Ai-generated handwritten letters, providing authenticity in a can.

This year we are likely to see this technology become more capable. In time, it will operate autonomously on our behalf.

Why? For many, the convenience of streamlining or outsourcing responses to the tide of emails, texts and calendar invitations that washes over us every day will be too tempting to ignore. Software makers will see opportunity in boosting their users’ productivity.

Why not? Machine learning is imperfect, and some AI scientists think it may soon “hit a wall”. Many consumers will be nervous about outsourcing their responses too much, and there will be ethical debate about whether it is a sign of disrespect. New privacy laws may also limit how closely third-party AI services can integrate with online utilities.