The challenges of rapidly testing our hypotheses about our AR app

Rapidly testing out a hypothesis for an Augmented Reality learning app is quite hard. My team divided our language learning app idea into two risky assumptions: 1) users will learn new syntax and vocabulary with the help of AR, 2) users will be able to implement the new knowledge and get feedback from a voice assistant. This week we focused on the first hypothesis.

First, learning is meant to be longitudinal. In order to rapidly test our hypothesis, we needed to come up with something snappier than a longitudinal study. Second, testing for the effectiveness of an AI bot that guides you in the AR space possibly is hard to do “cheaply” and quickly. Let me explain why.

What constitutes a wizard of oz testing is that the participant is unaware that a human is behind the decisionmaking process on behalf of the computer. If we were to do this in AR, it would be called collaborative AR, which in and of itself is a research area. So we opted for a concierge testing, where the participant was aware of the observer’s presence and influence.

We got the feedback that a bot that can tell what the things that it sees through the phone camera are is “useful.” But in what ways is it useful? Are our participants confounding excitement of something cool that talks to you with usefulness? Our answer, for now, is still perhaps. Stay tuned for more hypothesis testing.

Related posts:

JUAL POHON NATAL MURAH DI TAJINAN Pohon Natal adalah hari perayaan kelahiran Yesus Kristus, Sang Juru Selamat dunia. Momen tersebut sangatlah berarti dan membahagiakan bagi umat Kristiani. Tidak…

Introducing veganism. Helping a friend who is vegan. What to cook for a vegan. All vegan things and non-vegan things.

For this first spread, I changed the color of the black and white image to be true black and white and not the sepia version I had before. I also got rid of the color blocking and added in the hand…