Every year, Forrester puts together a list of 10 emerging technologies to watch. This year’s list was released in June, and in the most recent episode of our podcast, What the Dev?, we were able to sit down with Brian Hopkins, VP of Emerging Tech Portfolio at Forrester, about the list.
Here is an edited and abridged version of that conversation:
One of the things that stuck out to me in this year’s list is this idea that there’s been this shift from generative AI to agentic AI. Can you explain what agentic AI is and what the shift means?
Absolutely, the trend you’re talking about is a shift from focusing purely on generation of text using artificial intelligence to building AI agents that actually are capable of accomplishing actions on people’s behalf. When we think about an AI agent, we think about a piece of software that’s going to actually be able to take a general set of instructions, and be able to generate a visualization or access a database or trigger actions within another application. The most exciting thing we see right now is the shift towards actually using these next generation models in more of an action taking context.
As you wrote in the report, the rise of these AI agents is sort of giving way to a number of other emerging technologies on the list, including TuringBots, edge intelligence, autonomous mobility and extended reality. Can you briefly explain those other technologies and why AI agents are so important for their growth?
I think it’s important, before I get to those other technologies, to also explain the idea of an AI creating an intelligent agent. Earlier AIs that could go do things were narrow and constrained to a particular environment, using things like reinforcement learning. What we’re seeing today is taking the capabilities of large language models to break those instructions into specific steps and then go execute those steps with different tools.
When we think about that kind of design and how that might play out across a bunch of other emerging technologies in our list, a really interesting story starts to emerge. For example, one of the other emerging technologies we have is TuringBots, which we’ve been writing about since 2020. TuringBots are autonomous coding bots, and what we saw in 2020 was the ability, in theory, that given enough training data — like all the source code in a repository that you kept in a GitHub repository — you could train a machine learning algorithm to go write code based on that training data.
What we saw with generative AI in 2022 was that capability was dramatically accelerated, because before it gets compiled, software code is just text. So we saw that accelerate. When we first identified TuringBots as an emerging technology, we put it in the five to 10 years before we thought there would be benefit for most average enterprises. This year we moved it into the one to two year near term benefit horizon because of the acceleration that state of the art generative AI models are improving their ability to generate useful code.
We recognize that a TuringBot is itself an agent, and what we’re seeing is the use of agent methodologies to create, perhaps swarms of TuringBots that are operating in different developer capacities, from design to coding to testing to deployment.
And when we think about this, if we can produce software code with much lower human effort, we can iterate much more quickly. And if we can iterate on innovation ideas more quickly, we can get through the ones that aren’t good and produce the ones that are much, much faster. And we know that leads to an uptick in the pace of business changes.
You asked about the other emerging technologies, and I’ll be a little more brief. Edge intelligence is about using information that’s outside the data center or outside the cloud, outside of a centralized location, to process information and use that information to create action and intelligence. Prior to this year, it was mostly focused on things like computer vision. So you had a very narrow model trained to recognize certain kinds of objects, and it would go do that computer vision recognition well. But what you did with that recognition, frankly, then had to be programmed in some kind of heuristics or code.
What we’re beginning to see is — for instance, in the Apple Intelligence announcement but there’s others as well — how we are able to take agents that can do things, train them, and make them small enough to run on various edge devices. And then those edge environments, beyond just being able to perhaps converse in natural language with humans on the edges, can converse among themselves.
The example that we give is there is a vendor who is looking at creating augmented reality overlays in the next generation firefighting helmet, which is an effort being sponsored by Homeland Security. If we begin to think about putting agents in those helmets, then a lot of the communication those firefighters would have to do themselves could be handled by agents in each one of these helmets, looking at those augmented reality displays and making decisions about where different firefighting assets need to be placed to offload that from the need for human communication in a scenario like that.
So that’s an example of how generative AI is kind of serving as the foundation for agents, and those are then creating new innovation possibilities in edge intelligence. Same is true for autonomous mobility. We’re going to see these agents deployed in IoT environments, so that drones and robots can be a lot smarter in their communication with the surrounding environment. So we just see this whole idea of acceleration from generative AI creating a revolution in the ability to use it to do things, and then that’s moving into a bunch of other emerging technologies in our list and accelerating them.
A lot of the items on the list were AI related, but there are also three security items on it, so I kind of want to shift over to that side of things. So what have you been seeing in the security space that’s influenced the technologies that were on the list?
Actually, I’d like to answer that question the other way around. Why are things happening with the other emerging technologies that make security so important? This year when we did the research the idea dawned on us that those who achieve these future benefits are going to be the ones with the presence of mind to invest in security today.
I’m seeing this play out over and over as I talk to clients who are telling me stories that they’ve been meaning to invest in better IoT security for years, and they just don’t see the value in it, because it costs money and it’s complicated and it doesn’t have an immediate top line impact. IoT security is on our list and IoT security has been around as long as there have been devices to secure, you know, 30 years. Why is it there this year? We see an enormous amount happening in the space, and the reason for that is, very simply, all these AI tools that the enterprises are getting and figuring out how to use to their advantage are also tools available to the bad actors.
What’s happening is organizations invest in more devices, more smart connected things, and we’re essentially increasing the attack surface by which smart hackers with an army of very smart bots can launch attacks to get in an operational technology environment, your faxes, your printers, that thing that you haven’t updated the firmware for in 15 years is sitting in the corner in your office, connected to your network.
So IoT securities have really become really important, and there’s an awful lot going on in that space right now in terms of vendors and how they’re providing new capabilities for inventory and remediating all your IoT devices.
The other two are zero trust edge and quantum security. Zero trust edge is essentially a packaged set of technologies that give you a whole bunch of capabilities that combine networking and security into a cloud-based, as-a-service delivery model. So you get all the features of managed cloud services, and you get the ability to manage your security down at the network level, which means, according to principles of zero trust, you don’t have a firewall anymore. You inspect everything and trust nothing, and therefore you’re looking at all the packets going across your network.
The problem is that it requires firms to be pretty modern in their approach to cloud native software deployment and management, and a lot of firms are still pretty behind on that. There’s a lot of legacy devices out there, that fax machine sitting in the corner that doesn’t use modern protocols, modern security, doesn’t easily connect to this kind of agent-based zero trust edge architecture. It’s complicated, and the vendors are busy consolidating. So that’s why we think it’s going to take five more years before this really pays off. That doesn’t mean that today you can’t start working on it by making sure that you’re ready for a modern cloud-native way to manage both networking and security together. There’s a lot that needs to be done.
There’s also been a lot of hype around quantum computers for the last 10 years, and we think quantum computers are 10 to 15 years out from actually being able to threaten today’s best PKI encryption. So it’s easy to say, well, it’s 10 to 15 years out, I don’t need to do anything now. But nobody knows how fast quantum computers are going to advance. It could be a lot faster, could be five years. What you have to worry about is the attack of save now, decrypt later. You’ve got to start now implementing quantum safe algorithms to make sure your data is protected.
But the real reason we put it on the on the top 10 list this year is because implementing quantum safe algorithms and being able to rapidly change algorithms as quantum computers advance and new quantum safe algorithms are put forward, is part of a broader effort around cryptographic agility, and cryptographic agility has many benefits beyond protecting you from quantum attacks. New hacks are coming out all the time, so by looking at cryptographic agility solutions today in preparation for being ready for quantum attacks, you’re actually improving your whole security posture. There’s many benefits to starting now, which is why we put it in the top 10.
We covered a lot here today, so is there a takeaway that developers and leaders should come away with as they think about what to focus on in the next year?
You have to spread your investments out. Short term is easy, it gives me benefits that I can measure, my finance people like it. But you have to take some of those mid and long-term shots as well. And a lot of the long term things that we have will require big foundational investments to be ready.
I think corollary to that is with the speed of acceleration that we’re seeing happening primarily because of the advancements in AI today, we’re much less certain what the future is going to hold, and we’ll have much less time to deal with it. What that means is instead of saying here’s what the future is going to be, here’s our bet, you’re going to have to spread your bets out across a range of possible options. So you’re going to have to hedge your bets a little bit and use more of an options-based strategy to figure out where you spend your money, so that no matter which things break and go, we have a better chance of being ready for whatever happens.