The future is now: kill or cure robots, forensic face matching and social communications

Artificially intelligent machines are likely to play an increasing role in society in forthcoming decades. Do these machines need a sense of *ethics* to act effectively in human society? If I have a self-driving car, should it protect me at the cost of other people? Is it acceptable for a company to use AI to search for elaborate tax-avoidance schemes?
Should a computerised doctor be deciding on your treatment without consulting a human doctor? Should robot soldiers be allowed to autonomously choose targets?

Person identification at passport control relies on a comparison between a live traveller and their passport photograph. An important purpose of this is to detect “impostors”, who might be using a stolen or borrowed passport. The scale of this problem is currently unknown, however research attempting to estimate performance in this task suggests that face matching is a particularly challenging task, with up to 20% errors arising under highly optimised conditions. However, this is without considering additional factors that impact performance in these settings.

We will be talking about mindreading… that is, how human beings are able to ‘know’ other people by understanding what they may be thinking or feeling at any given time. Mindreading abilities are engaged so readily that it can be easy to forget just how critical they are to our daily lives – we are constantly working out the minds of other people, such as what they do/do not know, to ensure successful interactions! This talk will explore how social communication abilities change across the lifespan, and what may influence these changes, both positively and negatively, at different ages.