Jonathan Andrews spoke to Chris Crist , Chief Information Officer, Atlanta Hartsfield-Jackson Airport , about how the airport is using AI and the future challenges and benefits it can provide
How is your airport currently working with AI? We’re fortunate in that we have worked with Delta
We have a lot of data to show the time saved, and we’ve learned a lot from that and we’re expanding upon that capability now and wanting to implement it at all 40 international gates.
We’re working to first identify a partner and answer the questions of what we’re expecting, what we need, and the requirements. The intent is to make it agnostic, so that it’s not just with Delta anymore, so that all passengers, no matter who they are, no matter who they fly with, they can have the same experience.
We’re then going to expand upon that even further by working to install the biometric facial recognition capability for our employees for when they access secure areas. Right now, we use a badge and we use a PIN. We’ll still require a badge, but no more PIN. Instead, it’ll be the facial recognition which increases our security because no longer are we having to worry about PII [personally identifiable information] or someone getting that information.
What have been some early lessons from this so far? A lot of people in the beginning were against it because they were concerned about privacy. People can opt out if they want to but once they understand it, they realise that it’s not like that. We don’t store anything locally.
We’re just comparing your image that’s in the customs database, which is not stored locally with us, it’s with customs. Privacy was a challenge and people’s concern around the use of that technology, but we continue to educate people.
Looking at generative AI, what are the potential applications you’re seeing so far for airports? We just got the approval to move forward with the initial stages of creating an internal chat bot that will first focus on human resources. For example, staff can ask it what the policy is on wearing jeans, ask it about the policy on leave, or how much leave someone could carry over to the next year. That’s what we’re looking to incorporate on a very practical level, and that many other organisations could potentially see within months if they really wanted to push it.
During the first Airports AI Alliance session you talked about IT governance, and how you prioritise and evaluate investments and projects. Can you explain that in the context of AI deployment? We’re still working out the kinks of this new process and we’ve put a few new projects through our new workflow.
All business units now have an opportunity to have their say on what should be prioritised from a technology perspective. It’s good to have that debate, and it’s good for everyone to understand what the priorities are. We’re very much doing the same thing automatically with AI related projects but the difference with them, or data related projects, is that that actually starts with our innovation department.
Our innovation department, they report to me and for anything new, like AI related things, it’s going to them first. They’re evaluating up front, they have the innovative mindset, and they’re thinking, ‘Okay, well, we’re already doing these things with AI or data, so let’s talk about what we can do with that’. They’re doing the initial assessment, and if it seems worthwhile to move forward then we’ll run it through the formal process, like every other project, to prioritise it in terms of resources.
That’s really the only difference, in that we have an innovation team that’s tied to AI related things, and eventually AI will be so common that it won’t have to go through the innovation team.
Are you at the stage where you need or are developing AI policies for vendor agreements and data sharing? I would look forward to having that conversation with the group, because I like to better understand and learn from others what their perspective is. I had a conversation about this with some very smart people in the industry who are leading these things, very smart data scientists, very smart AI machine learning professionals, doctorate level machine learning professionals. I’m worried that people may be too concerned about the AI policy, the AI covenants, and guardrails.
I get it, but people may not realise that we already have something in place to guard against any concerns with new technology. We follow the National Institute Standards Technology Risk Management Framework
I don’t think we should be so concerned about generating something that is so specific to AI because do we have it for other technologies? Have we done that for hardware? We have it for other things but really, it’s tied to the risk management framework. You can’t create a policy on how to properly use AI, it’s more about how to appropriately use the tool. Do risk assessments on any new technology, no matter if it’s AI or anything else.
