Robots' Assistants?

While browsing reddit, I came across a link to an article about robot assistants. I thought to myself, “that would be cool to have,” then proceeded to scroll on. As I was scrolling, I realized what the title of the post actually meant. The article was not about robot assistants, but about robots’ assistants. The article was about a new type of job that is popping up as a result of countless Silicon Valley startups. This new job is to be the person who helps A.I. programs to do tasks that may currently be beyond their capabilities. The article explains how personal assistant programs, such as Facebook’s new “M”, have humans that help them when they encounter a request that they are not capable of fulfilling, or have audio that isn’t clear enough for the program to make out. If you haven’t heard of it, Facebook M is a new personal assistant feature that can do things like making reservations and making appointments.

AI assistants have been gaining popularity

These human “assistants” take the requests from the programs and handle them. In doing so, they build up a database of information about how to do these types of tasks. This allows the super complex algorithms to “teach” the programs how to do these things. At first, I would think that this meant that the human interpreters were making themselves obsolete by making the programs smarter. However, the article explains that as the programs learn to solve more problems, there are more problems that they’ll encounter. As the article puts it, “the more we learn, the more there is to learn”.

I found this article to be extremely interesting, as I find the limits of computers to be a very exciting topic. However, after thinking about it for a while, I came up with an interesting thought. I realized that now humans are working for computers. We have become their assistants, when just a few years ago using computers as assistants was all the rage. So humans are now being assisted by computers who are being assisted by humans, weird. Now, I understand that humans have always been working for computers to make them smarter, but there is something eerie about humans being passed the tasks that computers can’t handle, and then the computers learning from the ways the humans complete the jobs. I am not particularly worried about Skynet rising up and taking over, but it is extremely exciting to me that we have gotten so good at writing software that computers can teach themselves (with a little help of course).

On a slightly different note, one thing that the article mentioned that I found to be super weird was the existence of services that act as a fake girlfriend where users pay a monthly fee and can have text message conversations with a fake girlfriend who is actually just a large group of people who produce romantic replies. The company is called Invisible Girlfriend, in case you were wondering. I find this to be a strange product and I wouldn’t bet that it could have succeeded. But hey, whatever floats your boat. 

Article from New Scientist