The bots are coming, and if programmed correctly, can provide unique (and automated) opportunities for brands. Take @SmileVector for example - this account has been plastering a smile on images of celebrities using a neural network to identify and modify facial features.
New Zealand lecturer at the Victoria University of Wellington School of Design, Tom White, trained his Twitter bot to identify smiles (by inputting many, many examples).
From here the bot has been planting unsettling smiles on famous celebrities with impressive accuracy.
— smile vector (@smilevector) June 24, 2016
pic.twitter.com/8dnQEI71ye — smile vector (@smilevector) June 25, 2016
— smile vector (@smilevector) June 25, 2016
pic.twitter.com/FYD6BOpxOU — smile vector (@smilevector) June 27, 2016
Of course, such algorithms are still very much in their infancy, and when mixed with the variables chucked up by social media, can result in disaster.
A leading example is Microsoft’s TayBot which had been trained to react to user stimuli and messages. After a surprisingly quick period, the bot had been corrupted by web users, implanted with all manner of hate speech.