Yesterday Benedict Evans wrote about Disrupting Mobile. He noted how PCs disrupted mainframes and how mobile in turn disrupted PCs, and how in both waves of disruption a massive increase in the size of the ecosystem led to the increases in investment which meant the new paradigm prevailed. Looking forward, pretty soon a high percentage of everyone on the planet will have a smartphone making it hard to see where the next massive increase in ecosystem size will come from. Internet of Things and AR/VR use smartphone components and seem to be more of an extension of the smartphone ecosystem than a replacement.
At the very end of his post Benedict wonders whether AI might be the next wave. It could be ubiquitous and shift our attention away from mobile by making it less relevant, in much the way our phones have shifted our attention from our PCs.
I want to think about that for a second.
Today AI is forming the backbone of an increasing number of services (including a few in our portfolio) but it is working through mobile or the web, not making them irrelevant. For the next little while I see our phones becoming more useful and more central as innovators leverage their sensors and always-on nature to provide us better and better services, many of which will be driven by AI. In this future there’s no ‘iPhone moment’ that heralds in a new era.
Later on services are likely to get sufficiently intelligent that we need less sophisticated phones to interact with them. If interaction is primarily through speaking, and if processing is server side, then the smartphone quickly gets commoditised. In parallel, screens and devices will continue to fall in price leading us to own more of them and be less reliant on a primary smartphone. Extending this trend, computers will eventually become part of the fabric of the clothes we wear and the furniture in our houses and offices. At that point carrying a smartphone becomes less important and our attention has firmly shifted away from mobile.
This is more a vision of ubiquitous computing than AI per se, but AI will be a key enabler. User identification and authentication is another obvious challenge.