More and more services are transitioning to the internet and self-service. Data analytics and artificial intelligence help people in their daily activities. Now we see some services where people can get internet services that include human service components. Is it a new trend or is it too optimistic to think that a human can beat a machine in these services?
The New York Times wrote about a new phenomenon involving startups that offer human services over the internet (see In a self-serve world, start-ups find value in human helpers). The article details new companies that offer services like booking a complicated trip, finding a handyman or a plumber, and they use people to conduct part of the services. Those people can help to find and prioritize travel options, and also evaluate what kind of handyman or plumber fits your needs. These companies like Lola Travel and Happy Home Company say it is really easier to user to just a send a message describing what he or she needs, and they can help to find the best options.
Facebook announced they have a new service called M that is based on AI. When Facebook, for example, notices you travel somewhere, M can tell you the weather, give you tips, recommend places to eat, etc. But this created a discussion over whether this is really pure AI, or if people are working as part of the service. And it created an interesting scientific or even philosophical discussion into how to test if a service is really AI without human contributions. It would be a kind of Anti-Turing test to prove something is a pure AI. It is actually more complex to do than the original Turing test to prove a machine can behave like a human.
AI is becoming more important in many services, as I wrote earlier (Artificial Intelligence has come back). Now we start to see even legal services that utilize AI, like hirepeter.com, and the same happens for many white-collar jobs. We could say that services like Uber also use human resources, when drivers are people. But actually taxi services are heading towards self-driving cars.
Do we have two different trends happening at the same time - machines are doing more, but humans are needed more in internet services? Probably this is the case - when more and more services are going to the internet, some of them also need human resources. But is it a longer-term trend or an interim solution?
If we look at the travel services, it has been stipulated a long time that people are ready to pay for human services to book more complex trips or luxury holidays. Most probably this is the case for people who want to pay a premium. At the same time, does it make your booking much easier, if you just send a message to someone, who then replies with several options, and especially if this person doesn’t know all your preferences (airports, airlines, prices, hotels you like, or if your dates are flexible)?
I believe it can make the booking more complex than sites where you can choose all those options. Or if the system knows all your preferences, does the human help then to bring value add? Maybe not so much. And what are the real experiences, if you now must call to an Internet travel service call center? Your call goes to a call center somewhere on the other side of the world, and typically the call makes you so angry that you really would like to do anything online.
It is easy to believe human beings are needed in the transition phase, when some services go online and also for some premium services. But I’m very skeptical they are the long-term cost-effective scalable solution for mass-market internet services. The development of artificial intelligence, machine learning and deep learning is really fast. Machines develop more rapidly than human beings to handle these kinds of services. Human beings must find more demanding tasks to compete with against machines.