It’s here, again. The existential dread, the 10-point articles telling you to “sort your life out f'king finally, jeez”. On January 1st, prompt as ever, they arrived – emails promising to reveal the “top tech trends in 2018”, as if the universe has already been planned and no one thought to tell us before.
As usual, the same gushing misrepresentation of AI studded every list. Having dallied with AI in 2017, we’re suddenly now there guys: 2018 is the year of superintelligence. Drop your sandwich, it’s all ending; they’ve got us, they’re sticking stuff in our brains now, assume the foetal position etc.
The state of AI
It’s disappointing to still have to be dealing with these headlines, when the reality is so very much not the Philip K. Dick novel we keep being sold.
I mean, that’s not to deny the amazing progress AI research has made up to this point. Brain-computer interfaces which restore motor control, road-tested self-driving cars, image recognition networks that compete with experienced radiologists in diagnosing medical scans, computer listening tools that instantaneously translate speech into another language – this is the stuff of science fiction.
But AI isn’t just about blockbusters. In reality, a lot of AI application today is low-key and unglamorous. We’re talking automated small admin tasks, algorithms that show you stuff based on your past purchasing history, and awkwardly-named customer support chat bots who politely process your text-based frustration.
The case for "small AI"
Pretty boring, right? Actually, no. Small-scale AI is pretty damn exciting. We’re developing a completely new technology here; of course we’re going to be taking baby steps, of course not everything will be news-defining, and of course not everyone playing with AI has the resources or jowly zeal of Elon Musk.
Top-funded AI research should in no way eclipse the insane progress “grass roots” AI is making, from small start-up-led initiatives, to kids hacking neural networks together in their bedrooms, to people who have no tech background accessing and learning about AI tools.
So, this year we’re looking forward to all the banal successes: the invisible programs that simplify your back-end, the garage Kaggle contests, the reading groups discussing 'TensorFlow for Dummies' (yes, it’s actually a thing!). If it makes your life easier, and helps your customers and colleagues, it’s worth shouting about – no matter how small or seemingly simple on the surface.
Start positively and celebrate the little details. In a world blinded by the blockbuster, let’s make 2018 the year of “small AI”. And no, you don’t have to run a marathon/change your personality/plan your funeral by the end of this month.