Will you learn to trust artificial intelligence next year?

Best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews on Apple Podcasts or PodcastOne.

Taming the sometimes chaotic cloud computing arrangements and learning to love artificial intelligence, those are among the technology trends for 2023 identified by thinkers at Deloitte. For more on this and what else federal IT managers will have to deal with next year, the Federal Drive turned to Deloitte’s public sector chief technology officer, Scott Buchholz.

Interview transcript:

Tom Temin
An interesting list you have compiled here for the year ahead. Some of them are fairly basic technology, like taming multi cloud chaos, which I think everyone can relate to. But you also have immersive internets for the enterprise, and some more ethereal sounding things. So let’s begin with what do you think is the top thing that’s going to be happening in the year ahead from a technology standpoint?

        Join us May 2 at 2 p.m. ET for a discussion with agency and industry leaders on network modernization strategy, 5G, AI and plans for the future at the Naval Network Warfare Command and NASA, sponsored by T-Mobile. | Register now!

Scott Buchholz
Well, there’s sort of what I think and what I hope, how’s that, I think that part of what’s going on Tom, is people are realizing we all have collectively, tons of investments in cloud. And those are not just the infrastructure as a service investments that we’re making, but also Software as a Service increasingly, so. And part of what people are starting to find in that regard is we don’t necessarily have a consistent way to apply policies across them. And expecting individual humans to do individual configuration of individual things just doesn’t scale. And so I think, if you were to ask me, where I think people are going to be most focused next year, I think it’s that because we’re all focused on the problem in front of us.

Tom Temin
Right. And that kind of leads to the idea of automation of some of these high speed and high volume tasks, like making sure policies are consistent across cloud, otherwise, your cybersecurity breaks down. So that would be what Deloitte is calling opening up to AI learning to trust our AI colleagues, there’s been a lot of work done on AI a lot of investment in AI, in both the civilian and DoD sides. Are we seeing results yet? And are we seeing people embrace this in operation and not just in theory?

Scott Buchholz
We are starting to see people embrace AI and operations. And that is true across the government. And you know, globally around the world, when we look at other industries as well, I think what’s interesting about AI, is that in many cases, people’s mental model is often that artificial intelligence and computers are like calculators. And so what we’re looking for somehow, we expect them to be perfect, they’re going to be superhuman, in some ways, and perfect, not just as good or better than our best colleague. And one of the easiest ways to think about this is if you’ve ever sat behind the car with a teenage driver, and compare that to a self driving car, I’m not convinced that my teenage children are actually safer than the automation. But we worry more about the automation, because we’re expecting computers to be perfect, not just better than most humans. And the reason that that’s actually really important is because if we start to think of AI as a really wise colleague, who might just tap us on the shoulder and say, Hey, Tom, Hey, Scott, maybe you should take a second look at this thing, then we can actually deal with it more intelligently. Because we know how to deal with a wise colleague who’s often right, but sometimes wrong, as opposed to an infallible oracle that’s always right. And so I think governments will actually find that, the more they can think about wise, automated colleagues, the better that we’ll actually have in terms of AI adoption, because we won’t then be looking for perfection. And we know how to manage quirks and foibles of humans who are imperfect.

Tom Temin
Sure. So an example of application of that might be, say, an agency that has an adjudication function, and they’ve got 500,000 cases that are backlogged, maybe 400,000 of them are routine, you can almost rubber stamp them that 100,000 You got to flag those where you can then have a rational way of allocating the expensive commodity, which is human time and discretion.

Scott Buchholz
That’s right. Healthcare has actually been doing this for years, they’ve been automating in the insurance market, some of the back office approvals, and what the smartest way to do it tends to be is when the AI algorithm is highly confident that the answer is yes, you should be, for example, approved for a procedure, the AI is allowed to make that decision. But if the AI thinks the answer is no with a high degree of confidence, or the AI simply isn’t confident that it knows the right answer that gets routed to a human and so in that way, it’s not that you’re enabling the AI to make bad things happen faster, right? You’re basically letting the AI deal with the routine things that we’re confident are correct, but anything that looks squirrely or anytime you’re tempted to say no, it gets a second review.

Tom Temin
Maybe AI should be called IA which would be Intelligent Automation. Well, no charge.

Scott Buchholz
The best use cases are in fact for some version of that, yes.

Tom Temin
We’re speaking with Scott Buchholz, he’s chief technology officer for government in the public sector at Deloitte. And I wanted to ask you about this idea that you see for the future through the looking glass, you’re calling an immersive internet for the enterprise. Is this mainly in the training area? Or how do you see immersive types of environments applying on the federal government?

        Read more: Artificial Intelligence

Scott Buchholz
I think the easiest place for most of us to envision the future of things like virtual reality and, and other things in the near term is absolutely on training. And if you think about it anywhere, where you’re trying to put a person in danger, so that can be inspectors who are inspecting electrical equipment, or repair technicians, or service technicians, who are dealing with dangerous objects everywhere to people who are being put, you know, social workers in state governments who are actually being put in, you know, difficult situations in homes and trying to know how to react to things, that range of use cases, actually turns out to be people learn better, they retain better, and you get more value from the training than doing that same thing on a screen or with a piece of paper. Further out, it gets more interesting, because if you look around, we’re all trying to figure out how do we move beyond the little glass rectangles that we’ve been living with for 40 or 50 years, people are trying to figure out what is the future of the way we interact with technology, it’s probably more voice, more vision, more all sorts of things, and far less keyboards and mice and clicking and typing and that sort of thing.

Tom Temin
To some extent, we’re seeing that even in applications deployed to the public with facial recognition and sort of three dimensional ways of getting through traveling checkpoints. That’s not quite immersive technology, but it is a step beyond the interactions that we’ve had. So was that a decent example? Do you think?

Scott Buchholz
Yeah, I’ve even heard one of the one of the executives at CBP one time, say publicly, look, we get better security and better efficiency all at the same time through using some of these things. So that’s absolutely a great example.

Tom Temin
All right, and what about the tech workforce? I mean, you have said that there needs to be a reimagining of the tech workforce, with flexibility being built in there. Tell us more about that particular finding?

Scott Buchholz
Well, the first thing I’ll say is, you know, the government becomes an even more attractive employer, when economic times are hard. And so or, or uncertain. And so we’re clearly moving into an environment of more economic uncertainty, who knows how it’s gonna go. But that means that this is an opportunity for government to be a more attractive employer to people who have experience in areas that government wants to get into and cares about. There are two other things to think about, too, which is we’re also seeing more creative ways to get different people into the workforce, you know, different training programs, different levels of flexibility, different changes in terms of the requirements, and that’s helping attract people who wouldn’t ordinarily have gone into IT, or technology, you know, some of whom there are data science programs and development programs and other things where people who may not have four year college degrees are getting training to enable them to join the workforce productively, those are great sources of talent for government as well. And then I think the group we sometimes forget, is existing employees, because it’s easy to overlook. With the rate of change of technology, most of our skills are actually atrophying, and we really need to remember that we need to not just bring in new talent from the outside that has the experience we need. But retrain the people, we have to have the experience, we need them to move towards.

Tom Temin
All right, we could go on and on on all of these are pretty interesting, but I just wanted to not leave out the idea of getting back to some hardware here. Mainframe modernization hits its stride. What are you talking about here? Not really the iron itself. But the software systems that are run on mainframes?

Scott Buchholz
That’s right. I think the the thing we sometimes forget to acknowledge is the systems that run on our mainframes and have run for decades have actually served the mission well, for longer than some of us have been alive in some cases. And what we actually what turns out is when the mission evolves, when the mission changes, when expectations change, we need the flexibility to be able to move to meet those. And so what people are doing is they’re saying, look, the traditional technologies that let us manage things on the mainframe are not nearly as agile as the things that you know, the web developers of today use. So how do we use technology to help people increasingly move stuff off the mainframes? Componentize things on the mainframes? How do we take advantage of the AI? You know, and other things and the other technologies we’ve developed to chunk things up differently to solve the problem differently so that we can better take advantage of modern technologies so that we can meet the mission where it is not it where it was, you know, 20 years ago when we got started.

        Want to stay up to date with the latest federal news and information from all your devices? Download the revamped Federal News Network app

Tom Temin
As you stayed here, a funny thing happened on the road to obsolescence and people still keep these things going on. And really it’s just a calculating piece of machinery. And you might as well use it for what it’s best suited for.

Scott Buchholz
That’s absolutely right.

Tom Temin
Well, we could go on and on, but we are out of time. I do recommend looking at this report, if nothing else for the great artwork you’ve generated to go with it. And so blow it up on a big screen and you might get your find yourself stuck on the artwork.

Scott Buchholz
Thanks, Tom. We used AI to generate it. So it’s a little bit of taking our own medicine.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.