Artificial Intelligence: The days of siloed AI are (or should be) over

Today, most businesses have Artificial Intelligence (AI) strategies that are narrowly focused on assisting business processes. For instance, we might use this technology to automate processes, uncover design flaws, or pull insight from consumer feedback – but AI usually recommends rather than takes action, and is often siloed, not threaded across an entire operation. However, things are beginning to change, with AI shifting from assisting to action.

In fact, over the next decade, we expect to see entire AI agent ecosystems, large networks of interconnected AI agents and solutions that will push enterprises to think about their intelligence and automation strategy in a fundamentally different way. Agent ecosystems will let companies reinvent what and how they offer it.  Companies around the world recognise this opportunity, and according to Accenture´s Technology Vision 2024 report, 96% of executives agree that leveraging AI agent ecosystems will be a significant opportunity for their organisation in the next 3 years.

Nine in 10 (90%) Irish executives anticipate a medium to high impact to their organisation’s business processes in the next three years as a direct result of generative AI chatbots, with almost two in five (39%) reporting it as transformational change, according to Accenture’s annual Technology Vision Report for 2024.

A useful analogy for the progression of AI agents is the advancement of self-driving cars. For many years, drivers were entirely responsible for the operation of the vehicle (no AI). But then semi-automated systems like cruise control or lane assist came into play (AI that assists). After that, automated driving became available in limited conditions, and then fully self-driving cars requiring no driver at all (agents with increasing action). And if you extrapolate this trend, we can imagine a future with self-driving cars that all work together on the road (an ecosystem of agents). For cars, these advances have not come as precise step changes but as progress on a continuum. The evolution of AI agents will follow a similar trajectory.

The benefits for companies are substantial. Instead of using AI to optimise an isolated business process, agents could be extended to command entire chunks of the value chain. Today, AI can detect manufacturing flaws, but an agent ecosystem could enable true lights-out manufacturing that can increase safety for human coworkers. As AI evolves into agents, automated systems will make decisions and take actions on their own. They won’t just advise humans; they will act on humans’ behalf. AI will keep generating text, images, and insights, but agents will decide for themselves what to do with it. For instance, AI agents are already processing orders, yet agents could sell products and then get it to the customer’s door.

100% of Irish executives agree that the way we interact with data will change, from searching for information to asking questions and receiving direct advice and answers.

Four core capabilities to capture the value of agent ecosystems

But how can companies move forward? Through four core capabilities: access to real time data; reasoning through complex chains of thought; the creation of tools and aligning tech and the workforce.

1. Starting with access to real time data and services - when ChatGPT first launched, a common mistake people made was thinking the application was actively looking up information on the web. The reality is that it was trained on a vast amount of existing knowledge and drew on the relationships between that data to provide answers. For any tool to become a meaningful agent, it will need to combine the skills developed from a carefully cultivated historical record with the current information that comes from a rapidly expanding dataset of current events and knowledge.

2. The second step in the agent evolution is the ability to reason and think logically. Some may argue that AI has been doing complex tasks for some time now, such as playing chess or packing boxes. But these are relatively specific tasks, in which any deviation from the pre-trained instructions often results in failure. However, AI research is starting to break down barriers to machine reasoning.

Chain-of-thought prompting is an approach developed to help LLMs better understand steps in a complex task. Between chain-of-thought reasoning and plugins, AI has the potential to take on complex tasks by using both tighter logic and the abundance of digital tools available on the web. They imbue AI with the potential to navigate more uncertainty and with more solutions, opening greater opportunities for businesses.

3. The third dimension of agency we are seeing emerge is the ability for AI to develop tools for itself.  Researchers from Google, Stanford, and Princeton took a novel approach to how AI can create new, reusable tools to solve problems. They developed a closed-loop system comprised of two distinct AI models: the tool maker and the tool user. As the model receives requests, the “tool maker” creates functions that accomplish the objective, and then hands it off to the “tool user” – a separate, more lightweight AI model. Over time, the tool user can respond to requests that fit its growing set of tools, and the tool maker improves tools over time by learning from similar requests.

98% of Irish executives agree that AI agents will begin to collaborate with other agents to accomplish organisational tasks.

4. Finally, is the alignment between tech and the workforce. To achieve this, companies must understand that humans and agents are co-dependent. So, if they want to reinvent their AI strategy to tap into agent ecosystems, they need to reinvent their people strategy too, and reimagine the future of tech and talent together.

For instance, Accenture worked with Finnish Immigration Services to help the country attract the world’s most forward-thinking new businesses that could help build and  fund its high-quality public services. The companies worked with AI startup, boost.ai, to created one of the world’s first known AI-driven Virtual Agent Networks. This new “Living Service” consists of three AI-enabled virtual agents: one for immigration issues, another for business patents and registrations, and a third for tax-related queries. These virtual agents are powerful on their own. But together they offer a single, seamless experience to the end-user.

But it’s not just about new skills, it’s about ensuring that agents share our values and goals. In the era of agent ecosystems, most valuable employees will be those best equipped to set the guidelines for agents. Today, machines make up 43% of identities on enterprise networks and, as agents build their autonomy, humans must make and enforce the rules to ensure that they act for the betterment of the company and their people. A company’s level of trust in their autonomous agents will determine the value they can create. And the human talent is responsible for building that trust.

Before unleashing agents, humans need to embed rules, knowledge, and reasoning skills, and then rigorously test agents to ensure their readiness.

99% of Irish executives agree that generative AI will compel their organisation to modernise its technology architecture.

As agent ecosystems evolve, humans have two primary responsibilities to engender trust in semi-autonomous systems: building agent support systems and refining machine reasoning. So, the moment to prepare people to work in a collaborative way with new technologies is now.

From real-time information to reasoning, tool creation, and multiparty interaction, valuable agent breakthroughs are happening fast. This is why it is so critical to maintain focus on the evolution of the whole ecosystem – because as independently valuable as each of these developments are, their combination will spark a revolution in how we apply artificial intelligence. They will help build our future, and it’s our job to make sure they build a world we want to live in.  

Further findings from Accenture’s Technology Vision Report 2024 revealed:  Accenture´s Technology Vision 2024 report.