921 views1 Upvote5 Comments

VP of Product Management in Software, 11 - 50 employees
You have to make sure you've got the data as a foundation. I think people want to just jump into AI but first, you need to start with rules—try those out to see what works and what data you can get and then start to automate it from there.
1 1 Reply
Director of Security Operations in Finance (non-banking), 5,001 - 10,000 employees

Starting out with rules is a good thing.

Director of Security Operations in Finance (non-banking), 5,001 - 10,000 employees
Failing fast is an exceptional principle in my mind. That requires not only some level of use, but a willingness to revisit, reevaluate and review on a periodic basis. You need an understanding of the data sources and how to set whatever you want within your organization regarding those data sources, etc. There may be issues in terms of setting limits of utility until a technology has been proven out enough or until the AI has been proven out enough to say, "We're going to keep this in beta for this long and evaluate it this often before we take our foot off of the human level of interaction, etc., within the environment." Those are concepts that I'm hearing now that form some basic high-level, high-end principles regarding AI deployment. 

Another one is: understand the data sources from your vendors. Ask them the hard questions because they may not be adhering to the principles you're adhering to. And you may inadvertently cause bias within systems or do things that go against your ethical compass regarding AI, by bringing in vendors that don't share your beliefs or values.
1 Reply
Board Member, Advisor, Executive Coach in Software, Self-employed

An additional principle is one of transparency: you might understand all the data sources, but what if you don't have transparency in how the algorithm works? Even though you might answer some of those other things, it becomes difficult to know without understanding how it works. 

Transparency might be a root principle that runs through all this stuff. I think that's the macro one and other principles nest underneath it because in order to have the others, you have to start with a level of transparency. That is how I framed it in my mind.

President and National Managing Principal in Software, 501 - 1,000 employees
Understand not only where the data comes from, but also where it goes. Because if they can't clearly tell you where the data goes, you may want to question where it comes from. If they’re not able to tell you how the data is used, that means that there could be some red flags with respect to how or where it came from as well.

Content you might like




Non-production DBs (Dev, Training, QA, etc.)30%


1.3k views1 Upvote