As we delve deeper into corporate AI implementations, I’m really fascinated by how much it resembles your typical IT projects. Yeah, Generative AI systems sure sound special and sophisticated, but at the end of the day, they rely on some very traditional aspects. I’ve had conversations with several big companies about their various AI strategies and I’ve noticed that no matter what product you’re investing in, the success of the whole thing is directly related to your data strategy. If your data is a mess, there’s no way the AI is going to work its magic and make sense of it all.
I recently read a story about Microsoft’s Copilot bot spreading some controversial stuff, and that kind of thing really hammers home the point that we’re the ones responsible for the data quality, training, and security of our AI systems. It’s super cool to see Walmart’s My Assistant AI doing awesome work with employees’ inquiries about benefits, but it was only able to do that because it has a solid data architecture and tight security measures in place.
One of our clients, a big defense contractor, is figuring out how to revolutionize their knowledge management with AI. They’re still trying to figure out what data to load and how to best manage access to it all.
Then there’s the whole matter of security and access management. Remember back in the 1980s when IBM was all about RACF? Well, we need something similar for AI systems today since they also come with various layers of security and different levels of access for different users.
We also really need to focus on “prompt engineering” and system monitoring for these AI projects. We can’t design every single screen and menu like we used to in the old days – our users have to be able to tell the system what they need. We’ve built a “prompt library” for our Galileo system to make it more user-friendly, but it’s definitely a work in progress.
Another thing to think about is how to vet the vendors who are going to provide your company’s AI systems. Just because they can build a great AI model doesn’t mean they’re the right fit for your needs. It’s kind of like the new version of risk assessment for the IT world.
So, while the idea of AI is exciting, it’s going to take a lot of groundwork to get these AI projects to really take off. But, to me, this is an amazing time to be involved in technology and innovation!