A.I. will soon surpass our techniques for testing it as a technology, so it's time to look into how we can test and review cognition in order to grow as a discipline.
How can we test something that is constantly learning, changing, and might be hiding things from us? Artificial Intelligence (AI) technology is increasing at an exponential rate, meaning that soon most of us are going to at least have some experience in either testing an AI product, working with AI tools, or even being assisted by an AI “colleague.”
That’s where psychology comes in. As a clinical neuropsychologist I had to “test” people, who are also learning, changing, and sometimes lying. I think a lot of the same principles there can be applied to AI testing, and so I’ve set out to recon, understand and chart the methods and tools we might need to learn very soon.
In this talk I will walk through the questions and challenges we have today, as well as the work done so far in AI testing. Expanding on those foundations we will take a look at psychology and how the same principles used there can aid us in testing and working with AI and further evolutions of our technology. My goal is to show a new direction of growth in our testing discipline, one that acknowledges and embraces how human we will need to be.