Fears Growing On Power Of AI
For months, Gmail users have had live demonstrations of Artificial Intelligence (AI).
Its Smart Replay functionality shows an uncanny ability to read and understand incoming mails and to suggest possible replies.
For instance, if a friend or colleague emails ‘took a turn for the worst so will not make it,’ Gmail will offer ‘Oh no, feel better,’ ‘we will miss you’ or ‘sorry to hear that.’
Click your pick and a pre-filled reply window pops up for further text to be added or to send as is.
We might wonder if such a feature could be abused.
The replies users choose could tell something about them, like those personality tests in magazines.
Could this, in turn, inform approaches to feeding adverts back?
Allegedly, using AI in such a way has got Cambridge Analytica into hot water.
The data science firm was engaged by the Trump team to bombard voters with campaign messages during the 2016 US election.
Now, Facebook has suspended Cambridge Analytica from using its platform because, it says, the company harvested content from 50 million Facebook users illicitly.
It claims they analysed the posts using AI techniques to work out ‘which buttons to press’ in individually tailored election campaign messages for each voter,
CA denies wrongdoing and says the information used was obtained legally.
No matter which outfit in this row is correct, the real message here is the power of AI to examine what we say and work out how we might be manipulated.
If our thoughts can be purloined and used to influence us into doing something we otherwise might not, we need to guard against it, individually and collectively – and fast!
