Creating a test group to test AI systems, analytics, recommender systems, etc

All, it’s been quite a while since I’ve visited “The Club”; hope you all are doing well. I have a challenge in front of me - I need to create a group of professional testers that specialize in testing AI systems, recommender systems, and very sci-fi analytics with our data scientists. Have any of you built similar groups from scratch? Can you offer me some advice? I have both functional and automated test experience and have built many groups covering those specialties over the years, but this is somewhat out of my area of expertise - do we have any experts here that can tell us what it takes to build this expertise?

1 Like

Good luck in finding testers who drink that cool-aid. We are a skeptical bunch by nature as you well know. If I were you, I would be tweaking my linkedin profile significantly, and then using that as a blogging platform and hooking in anyone who give you likes and comments.

You will probably use a simple “AI engine” to suggest tags for your linkedin blog posts, and blog about how you hacked the linkedin tagging system as a route? Yeah, sorry, a bit out there, I’ve never seen a recommendation system deliver me anything that I wanted badly enough to buy it, in 20 years of being online. So I would struggle technically there, even though I think it’s a brilliant discipline in principle.

Best of luck Linda,
good to see machine learning/intelligence represented.

Was just thinking about my own bias of recommendation systems and how the tool is not a new tool, but how some people never learn how to use it well. Maybe that’s a topic, because it’s most often the deployment of these tools when it occurs in a environment outside of an actual “store” not being conducive to the user’s learning process. In my case I use these tools to get me to the correct “aisle” in the store for example, but no further. Because the AI has never learned that my purchase or choice is often using a metric it does not understand, nor can. Namely “play and differentiation”.

Would love to see blog posts covering more ways to grasp how and why, the machine has raced on ahead of the human in terms of learning. All the best.

I get the prejudices and even share them, but my reality is I need to add another group to my existing functional and automation teams that focuses specifically on testing AI and recommender systems, as well as testing analytics (which is the easiest part if you are into numbers). What my company does is not store-based; we enable people to give to churches and charities using their phone. Our company does much more than just enable a transaction from here to there; we are heavily into the psychology of giving and the staff that work on these things (in R&D) are PhDs in the “good meaning” of that word. The stuff they’re doing is fascinating and I need some brilliant people to help them get practical and test out their theories and systems. Ethics and integrity are a big thing here, so there’s a ton of discussion about what is actually useful and desirable vs intrusive and/or annoying. If I had the luxury of working as an individual contributor I’d want the gig myself, it’s exciting, new and I don’t know it well, but I’m a director and don’t get let loose often to test lately. But that’s what I did for an embarrassing number of years. I’m hoping someone out there has had to build a competency for this in their own companies and can help me not make too many mistakes as we expand our capabilities.