AI: It’s What Staff Craves!

If you’ve seen Idiocracy (side note: I don’t recommend the movie; I’m using a phrase from it that we’ve all heard. The movie itself was a comedy that was never meant to be family friendly when filmed, and changes in social norms since its release in 2006 make it less palatable today) , you are familiar with the address phrase I used here. This is almost religiously repeated to justify why I buy/use/drink Brawndo, an energy drink.

One scene includes people telling the star that plants need a Brawndo – “They contain electrolytes…that’s what plants crave!” This is repeated over and over until he asks, “What are electrolytes?” After some stuttering and failing to answer, one said, “That’s what they make Browno out of.” Another person inevitably says, “This is what plants crave.”

Now, replace “AI” with “electrolytes”. This is where we appear. Ask the vendor what algorithms they use, where their training data comes from, and the difference in results between test bookings and the real world. Most will end up saying too much, not answering your questions and ending up with, “This is what employees crave!” or some like that.

A few companies will put all of that out there, assuming you can analyze the information and be familiar with the AI ​​training algorithms they use or the algorithms they use to categorize the input against the data set. These companies are great because (a) I know What they actually do goes beyond “having AI/machine learning, that’s what employees crave,” and (b) Expect That you are aware of the data you have requested or are able to search for it in this age of exclamation that has limitless information at your fingertips.

Don’t let sellers get away with just repeating the phrase “contains electrolytes” over and over. Artificial intelligence and machine learning are a huge help in some areas, but they are not a panacea yet, success rates vary greatly whether we are talking about problem areas, data sets, algorithms, training method, etc., have them tell you the details. Where does the data come from and how big is this data set. How much was booked for the test and if it was booked at random. Ask them what they hope AI will achieve, and then ask them how that relates to human interaction. Most AI today still evaluates or replicates human interactions with something – most AI in IT in particular. What algorithms are used for validation? Finally, be sure to ask for realistic results. I know people in data science and someone has confidently told me, “My models are 95% accurate.” When I finally delved into this claim (because my experience indicates that real-world AI/Machine Learning use is hardly ever accurate), it turns out that this claim was true…for the reserved test data. Therefore, the same data set used in training obtained the same results in the reserved subset. This is a good thing, but it doesn’t indicate how said algorithm works in the wild with unfiltered data. The biggest problem is the shift from test data to real world data.

In fact, in IT, we’re a bit better off than most spaces grappling with AI/ML, because often there’s no difference between test data and IT training. The network security resource will test the actual traffic – because they have access to huge amounts of it and limit the iterations to use what is at hand and what the system is trained for.

And it kept swinging. The current iteration of IT is the opposite of the dark comic future shown in Idiocracy. You all set and keep systems running at increasingly fast paces; Sometimes remote due to COVID-19 or other work related issues. Don’t stop creating success for your co-workers and customers and take a moment to force sellers to explain what they’re doing with AI. If you’re a generalist like me, ask via email, so you can research the answers and make sure you understand the details.

Leave a Comment