We just entered an AI-First World
AI is almost everywhere, from your mobile keyboard software to voice recognition, personal assistants, and translation services. It decides what you see online, which songs you should listen to, and what you potentially want to buy. Machine learning became intertwined into our daily lives faster than any previous technologies. In 2015 only a handful of industry insiders, visionaries and nerds had AI under the radar, today is all over the news.
AI is here, and it will only get bigger — embracing it will be a huge competitive advantage, refusing it will most likely be professional suicide. AI and automation will reduce costs, improve efficiencies and provide a strong competitive advantage over late adopters. Robots will do what we do, but better and cheaper.
The AI platform is less important than the data sitting underneath it. That is why large companies such as Google released their AI software as open source projects. The value behind AI is the model data behind the standard algorithms. Google, Amazon and Microsoft have a strong advantage to collect, build and store massive amounts of model data. Early adopters will collect more data, which in turn will create better models and services, whilst late adopters will struggle to build up similar amounts of data.
One good example is image recognition. It generally uses Convolutional Neural Networks to train and recognize images on a large data-set. The more images to train the network the better it gets. Anyone can setup an image recognition system using Google’s TensorFlow as all the software required is open sourced. The differentiation factor between a self made image recognition system and a commercial one such as ClarifAI is the size of the training data and model behind it.
OpenAI is also helping to propel the progress of Automation in the real world. Simulations made in a virtual world can now be ported into the real world with little friction. It means that most automations made within the matrix will translate into real applications. Imagine teaching a robot to peel a potato in a simulation and seconds later having a robotic arm peeling hundreds of potatoes an hour — For less than minimum wage.
AI will lead to massive job displacement and potentially dangerously high rates of unemployment — It is a fact. The speed AI is progressing is faster and more far reaching than most experts predicted. Self driving cars and trucks will end millions of jobs in a matter of years. Self service kiosks already replaced countless jobs in fast food chains and OCR software replaced even more data entry jobs.
Getting into AI is still quite complex. There are a number of platforms, frameworks, types of neural networks and applications. Business leaders need to closely watch the space. It is important to work on AI applications that can provide a competitive advantage by leveraging current data and processes.
There are two applications with the highest return on investment at the moment. The first one is to leverage existing data to increase revenue per customer. The second one is more worrying, as it involves creating software automation to replace human labor.
It is important to embrace AI and the coming inevitable structural changes in the fabric of society. The real challenge will be to achieve growth without creating unemployment and unbearable inequalities. There are a number of potential solutions such as Universal Basic Income, robotic taxation and human labor tax breaks — but this is a much more complex discussion that the whole society should have, very soon.
I, For One, Welcome Our New Robot Overlords.