I've been trying to write this post for a while but am finally just throwing my hands up about having an well-done oeuvre to just get the thing done.
Used to be that computer science was like "automate all the things." More automated, more better. Bates (1990) was all like wait a minute here, there are some things it makes sense to hand off and others it makes sense for the human to do. People do some things faster. People learn and explore and think by doing. People need to control certain things in their environment. But other things are a hassle or can be easily done by a computer. What you don't want to do is to make the effort of supervising the automation so arduous that you're trading one hassle for another.
For quite a few years, there has been an area of research called "mixed initiative" that looks specifically at things like virtual assistants and automating where it makes sense without overburdening the user. As I was dabbling in this area a couple of years ago, I read some articles. It seemed weird to me, though, because I think most knowledge workers my age or younger probably don't know how to work with a living human assistant. I have never worked anywhere with a secretary who offloaded work from me. Never worked somewhere with someone to help me schedule meetings, type out correspondence, format articles, do my travel stuff, etc. I have been on teams with deliverables that were sent through an editor - but that was like a special technical writer. I suppose I would have to negotiate with an assistant what I would want him or her to do and then accept (within boundaries) that they might do things differently than I do. I would have to train them. Should I expect more of a virtual assistant?
All of this is in the back of my head when I started following the links.
So what do they mean by virtual assistants - they're hot, but what are they doing and do they work?
- Meekan is, apparently, a bot that takes an informal request within Slack and negotiates with other calendars to make an appointment.
- x.ai is similar but you cc Amy (a bot, but I like that she has a name), and she takes on the negotiation for you.
Project/Team Management (loosely construed)
- Howdy will get feedback from team members and also take lunch orders. Seems sort of like some things I saw baked into Basecamp when I saw a demo. It's in Slack, too.
- Awesome helps manage teams on Slack.
Travel, Shopping, ...
- Assist does a few different things like travel and shopping.
General but often operating a device
- Amazon Alexa
- Google Now (sorta)
- Facebook M
A lot of us don't want to talk to our assistant, but to text them. One of the articles pointed to this.
When I talked to engineers back in the day about their personal information management, there were a lot of things they were doing themselves that it just seemed like they should be able to offload to someone who is paid less (Pikas, 2007). Likewise, I was talking to a very senior scientist who was spending hours trying to get his publications to be right on the external site. Even though statements are routinely made to the contrary, it seems like work is pushed off from overhead/enterprise/admin to the actual mission people - the scientists and engineers - in an attempt to lower overhead. It pushes money around, sure, but it doesn't solve the goal. So here's an idea, if we really, really, really aren't going to bring back more overhead/enterprise/admin folks, are there bots we can build in to our systems to ease the load?
If Slackbot watches you and asks you personal questions: isn't that cute. If Microsoft does: evil, die, kill with fire. If your employer does: yuck?
Bates, M. J. (1990). Where should the person stop and the information search interface start. Information Processing & Management, 26(5), 575-591. doi:10.1016/0306-4573(90)90103-9
Pikas, C. K. (2007). Personal Information Management Strategies and Tactics used by Senior Engineers. Proceedings of the Annual Meeting of the American Society for Information Science and Technology, Milwaukee, WI. , 44 paper 14.