Last week, market research firm Forrester held its Technology and Innovation Summit North America, and it’s nearly an understatement to say AI stole the show—especially AI agents.
Broadly, many expect the market for AI agents to grow significantly from today, driven by demand for automation and increased AI adoption, with edge AI and applications such as customer service chatbots, autonomous vehicles, robotics, financial trading systems, and more. Research from MarketsandMarkets expects the AI agents market to hit $47 billion by 2030, up from $5 billion today.
Brian Hopkins, Forrester’s vice president of emerging technology portfolio, discussed the top 10 emerging technologies that will be crucial for CxOs to master over the next 1 to 5-plus years if they want to enable innovation and digital transformation.
Generative AI for Language and Visuals
The accelerating advancements in both generative AI for language and generative AI for visual content are driving the creation of multimodal AI applications, and the use cases for both are driving adoption. There has been a dramatic acceleration in the capabilities of these language models, as evidenced by the advancements from GPT-3 to GPT-4.
According to Hopkins, while Generative AI for language has many established use cases, marketing is the primary use case for generative AI for visuals. “Marketers have the most need to create visual content at scale that is increasingly personalized and tailored for their clients or customers,” Hopkins said.
One of the key trends here will be the convergence of the two types of generative AI into multimodal applications, Hopkins said. Hopkins cited Microsoft’s VASA-1, which uses generative AI for visual content to create realistic images of heads and then uses generative AI for language to generate synchronized text-to-speech output. Also, OpenAI’s ChatGPT 4.0 Vision project allows users to input visual content as prompts to generate text output.
Autonomous AI Agents
According to Hopkins, AI agents are general AI systems that can act invisibly on behalf of organizations or individuals. They are trained to perform tasks, make decisions and interact autonomously. They are often based on large language models that have been specially trained and tuned, and they are unlocking new use cases beyond traditional enterprise applications, such as better customer self-service, tailored marketing and consumer “digital doubles.” They are accelerating the development of other emerging technologies discussed, like Turing bots for software development.
“They are enabling a future where interaction with technology will be primarily through natural language and large language models rather than traditional applications and interfaces,” Hopkins said.
It’s already happening. At Dreamforce this week, Salesforce co-founder and CEO Marc Benioff said autonomous agents will provide value where generative AI “copilots” have not. Next month, Salesforce will release Agentforce, AI agents that promise to help staff work more efficiently in sales, marketing, service and more.
TuringBots
TuringBots refer to software agents that can automate various aspects of the software development lifecycle, such as generating code, testing code, deploying code, documenting code and architecting software objects, Hopkins explained.
TuringBots are themselves agents, part of so-called “agent swarm” frameworks that can instantiate multiple agents to handle different software development tasks. This allows developers to quickly work within the framework and supervise/direct an “army of agents” from natural language requirements to operational production code. The use of TuringBots and agent swarms is described as the “hottest trend in software development around AI today” said Hopkins.
Edge Intelligence
As described by Hopkins, edge intelligence refers to technologies that help to capture data, embed AI inferencing on devices and connect devices with real-time insights across application, device and communication ecosystems. Hopkins highlighted a few key reasons why edge intelligence is essential now: It enables technologies to run AI models on devices rather than relying solely on the cloud. This can provide benefits like faster response times and reduced bandwidth needs. When combined with AI agents, edge intelligence unlocks a wide range of use cases and opportunities for innovation, especially in consumer-facing applications. Apple’s ecosystem is an example.
The Apple Intelligence example shows how companies optimize foundational AI models to run efficiently on their own devices and ecosystems, creating agent-ready environments for developers, Hopkins said. Hopkins warned that overcoming legacy technology barriers is a crucial challenge for CxOs, but the potential benefits drive increased investment and innovation.
According to Grand View Research, the global edge AI market hit about $15 billion in 2022 and is expected to reach about $66 billion by 2030.
Extended Reality (XR)
XR refers to technologies that combine the physical and digital worlds, such as augmented reality (AR). According to Hopkins, the importance of XR is seen in AR applications, such as firefighting helmets, which can provide valuable information overlays to workers in the field and help them focus on their immediate tasks. When XR technologies are combined with AI agents, Hopkins said, it enables new capabilities where agents can coordinate, assess and prioritize what the worker is seeing and provide proactive guidance.
In the future, AI agents installed in intelligent buildings could also communicate with and provide information to firefighters wearing AR-enabled helmets, helping to direct them to the correct locations. Hopkins says this integration of XR with AI agents that can communicate using natural language may unlock enormous innovation and capability.
Hopkins said XR opens new possibilities for enhancing worker productivity, situational awareness and real-time decision-making, especially in industrial and public safety.
Autonomous Mobility
Autonomous mobility combines hardware, software, data and communications infrastructure around autonomous vehicles and robots. In the near term, the focus and use cases for autonomous mobility are in transporting goods and materials, managing large physical spaces and public transportation – not consumer self-driving cars, Hopkins said.
Hopkins used the Ursa AI urban trash robot as an example of how autonomous mobility technologies can be applied to automate and optimize tasks in controlled environments. He said that combining autonomous mobility technologies with AI agents unlocks new possibilities for coordination and optimization. For example, autonomous robot agents could communicate with smart city infrastructure agents to improve efficiency.
Hopkins said autonomous mobility enables new automated capabilities in transportation, logistics and management of physical spaces – and the integration with AI agents is seen as a critical driver for unlocking further innovation and optimization in these domains.
IoT Security
The number of IoT-connected devices, especially in operational environments, is expanding dramatically, as is the addressable attack surface for threat actors. As these IoT devices become more connected to enterprise networks, it increases the risk of breaches, data theft and other cyber attacks if the devices are not adequately secured, Hopkins said.
Hopkins noted that traditional “dumb” botnets made up of IoT devices are growing sophisticated, and the potential to leverage AI capabilities in these botnets makes them potentially much more dangerous. Hopkins sees investing in IoT security solutions as a near-term priority.
Quantum Security
Quantum computers, while not yet a practical threat, are expected to eventually be able to break current public-key cryptography (PKI) within the next ten or so years. And quantum computing will require quantum security. Quantum security technologies like quantum-safe algorithms and quantum key distribution are seen as ways to future-proof security and prepare for the eventual rise of quantum computing.
Hopkins noted that quantum security provides benefits beyond just protecting against quantum attacks. It also enables “cryptographic agility” to rapidly identify new threats and swap out encryption algorithms.
Hopkins said that preparing for this transition is crucial for organizations.
