If, like me, you do a reasonable amount of traveling, there's a pretty good chance that, also like me, you use Expensify to track and report your travel expenses. It's a pretty neat app. You use your phone to snap a photo of an expense receipt, which is then automagically "smartscanned" by the app. Expensify grabs all the needed info from the receipt and can even classify the type of expense and plunk it on a new expense report for you.
Imagine my surprise when I just recently learned that Expensify uses humans as backup when the technology is not able to read the receipt completely. Which means that, despite what you might imagine goes on behind the scenes, there may be actual humans reading your receipts.
How to start an AI startup— Gregory Koberger (@gkoberger) March 1, 2016
1. Hire a bunch of minimum wage humans to pretend to be AI pretending to be human
2. Wait for AI to be invented
Apparently, this practice is quite common. It's even extending to the world of self driving cars: Phantom Auto is a startup that provides remote control human intervention to self driving vehicles to help with "edge cases" where the technology doesn't know what to do. Very reassuring.
This is not surprising to me. Despite dire warnings about robots taking our jobs, and optimistic claims about cars driving us around in the very near future, it seems to me that AI has quite a way to go before it's able to do these things safely and accurately. Case in point, I was just texting a friend, asking about their 4th of July plans when I noticed Siri's suggestions via predictive text:
You might be wondering what any of this has to do with managing your Lotus Notes application migration project. Well, when we were first working on Teamstudio Adviser at a conceptual level, we wanted to create a product that not only would collect and present all the information you need to make decisions about what to do with different applications, but would actually make suggestions for you.
Organizations that are considering moving off the Notes/Domino platform often don't know how many Notes applications they have, and what the business impact and cost of moving them to a new application platform would be. The idea was to build a product that would gather together all the information that would be needed, and then actually make a recommendation for what to do with each application (archive, migrate, retire, etc). Since a typical organization could have thousands or even tens of thousands of applications, that could cut the time and cost of a migration project significantly.
That sounded a bit like an AI problem to me, and, since part of my role at Teamstudio is to come up with stupid ideas that make the development team roll their eyes, I suggested using IBM Watson to solve that problem. That suggestion had the desired effect, and roll their eyes they did.
Of course, as usual, they were right. The problem is not really complex enough, nor would we have access to the huge amount of training data that you would need to set up even a simple statistical classifier, never mind a full blown AI solution like Watson. Our actual solution is, of course, not something I want to detail here, but it's a fairly straightforward decision tree mechanism which turns out to be very effective. (And unlike self-driving cars, it doesn't need any human intervention.)
The problem we are trying to solve is not the sort of thing that, because of its complexity, can only be resolved by a human brain. The thing that makes it difficult is managing the volume of data, not the complexity of the problem.
Which is a good thing, based on my assessment of the state of the art with respect to AI.
Teamstudio Adviser and our other migration tools are being used every day by companies around the world to help them manage their Lotus Notes application migration projects. If you'd like to chat about how they can help you, just click below to start a conversation. You'll get to speak to a real human, certainly not a human pretending to be a machine pretending to be a human.