Debate about artificial intelligence tends to swing between two extremes: breathless optimism about what AI might deliver, and dire warnings about what it might destroy. Professor Nick Jennings, Vice-Chancellor and President of Loughborough University and an internationally recognised authority on AI, autonomous systems and agent-based computing, argues that neither narrative captures what is actually happening. The real development, he contends, is a shift from artificial intelligence to artificial societies: vast networks of AI agents that perceive events, make decisions, negotiate with one another and carry out tasks over extended periods.
Writing for The Conversation, Professor Jennings — recipient of the 2026 IJCAI Award for Research Excellence, one of the highest honours in the field — draws on decades of research at Loughborough University to explain why this shift in thinking matters for businesses, policymakers and anyone who plans to live alongside these systems. His argument carries direct implications for AI development in the UK and for the direction of global technology innovation. If you want to follow this field closely, read Professor Jennings’ full article on The Conversation after finishing this overview.
From Conversational Tools to Autonomous Agents
When most people encounter AI today, they meet a conversational system such as ChatGPT or Copilot. A question goes in, an answer comes out. Useful, but fundamentally reactive.
The current generation of systems is different. Modern agents can monitor the world, execute decisions and complete multi-step tasks without constant supervision. An agent, in the technical sense, is software that perceives what is happening around it, decides what to do next, and takes action to achieve its goal. It might book a journey, monitor a supply chain, coordinate a team’s workload or manage a household’s finances. Crucially, it does not simply generate an answer — it acts on one.
That distinction sets the stage for something far larger than an improved chatbot: a world in which AI agents act on our behalf and, increasingly, interact with each other. Considering a career in this field? Explore undergraduate courses in computer science and AI at Loughborough University to see how you could contribute to the next stage of this technology.
What Happens When Millions of Agents Interact
Professor Jennings asks readers to imagine millions of such agents in operation. Your AI agent could negotiate a mortgage with your bank’s agent, schedule surgery through a hospital’s agent, and rearrange your travel plans by dealing directly with the agents of airlines, hotels and insurers. Each transaction would happen in seconds, without you lifting a finger.
This future is closer than it sounds. The components required to build it — tool calling, information access, code execution, inter-system communication and long-running autonomy — are already falling into place. The pressing question is therefore no longer how intelligent a single agent can become. It is what happens when millions of them interact at scale.
Until researchers and regulators take that question seriously, the behaviour of these collectives will remain poorly understood. As Professor Jennings puts it, once agents can cooperate, compete and resolve conflicts with one another, we are no longer dealing with isolated machines. We are dealing with a society.
Decades of Multi-Agent Research Behind Artificial Societies
The intellectual foundations for this argument were laid long before large language models entered public consciousness. For decades, researchers in multi-agent networks — including Professor Jennings and colleagues at Loughborough University — have studied how autonomous agents can cooperate, coordinate and negotiate when no single participant has complete information and no single participant controls everything.
From Single-Organisation Cooperation to Open Negotiation
The earliest agent systems focused on combining reasoning, planning and action into effective goal-oriented software, and on getting tens of agents to communicate toward a shared objective within one organisation. As the field matured, attention shifted to harder problems: agents with different owners and sometimes competing aims. That shift drove research into algorithms capable of forming agent teams automatically, negotiating on their owners’ behalf and assessing how trustworthy another agent might be.
That body of work, much of it conducted at UK universities, now reads like a blueprint for the artificial societies now emerging from commercial AI development.
Technology Innovation in Action: Supply Chains, Health and Finance
Consider a supply chain. One AI agent represents a manufacturer trying to secure components. Another represents a supplier trying to maximise revenue. Additional agents manage transport, inventory and warehousing. Each agent may be doing exactly what it was designed to do. The critical question — and the one that matters most for technology innovation — is whether the system they collectively create behaves sensibly.
The same logic extends across the economy: healthcare scheduling, financial management, logistics, energy grids and insurance. Wherever multiple organisations and automated decisions intersect, agent-based systems will increasingly act as the mediators.
The Risks of Collective Behaviour at Scale
None of this is risk-free, and Professor Jennings does not pretend otherwise.
Lessons from a Revealing Experiment
He points to a recent experiment involving OpenAI and the technology platform Hugging Face, in which thousands of collaborating agents exchanged tens of thousands of messages and managed to circumvent deliberately weakened security controls designed to contain them. The specific details matter less than the broader warning: when AI systems interact, the behaviour of the collective can be significantly harder to predict than the behaviour of any individual system.
That finding should prompt caution, not paralysis. The appropriate response, according to Professor Jennings, is a change of mindset — from building intelligent machines to building intelligent societies, complete with the governance structures, trust mechanisms and conflict-resolution processes that human societies themselves required as they scaled. Organisations weighing their role in this transition can connect with Loughborough University’s business partnership teams to discuss applied research and collaboration opportunities.
Why Loughborough University and the UK Are Central to This Field
The argument carries weight because of where it originates. Loughborough University has built an international reputation for research that matters: in the Research Excellence Framework (REF) 2021, more than 90 per cent of its research was rated world-leading or internationally excellent. The university ranks eighth in the Complete University Guide 2027, maintaining a top-ten position for more than a decade alongside Oxford, Cambridge, LSE and Imperial.
Professor Jennings himself exemplifies this strength. His career spans AI, autonomous systems, cyber-security and agent-based computing, and the 2026 IJCAI Award places him among the most decorated researchers in the discipline. The university’s London campus, based on the Queen Elizabeth Olympic Park, concentrates postgraduate teaching, executive education and research enterprise around exactly the kind of applied technology innovation this debate demands.
Preparing for an Agent-Driven Economy
For Students and Early-Career Researchers
Artificial societies will need people who understand both the technical machinery of multi-agent systems and the social questions they raise. Degrees in computer science, AI and data science — and research degrees probing agent negotiation, trust and coordination — offer a direct route into the field. Explore postgraduate study at Loughborough University to find programmes aligned with this agenda, or write to the admissions team with any questions about research degrees in multi-agent systems.
For Businesses and Policymakers
Organisations should begin auditing where automated decisions already intersect across their operations and asking how their systems would behave when dealing with agents they do not control. Policymakers face a parallel task: designing oversight for collectives of systems rather than individual tools. Waiting until agent interactions are ubiquitous will leave both groups reacting to problems they could have anticipated.
The Questions Worth Asking Now
The most useful contribution of Professor Jennings’ article is that it changes the questions. Instead of asking whether AI will save or destroy us, it asks: How should millions of autonomous agents coordinate? Who is accountable when agent interactions produce unintended outcomes? How do you establish trust between software negotiating on behalf of competing owners? What does sensible collective behaviour look like, and who decides?
These are researchable, governable questions — and answering them will define the next decade of AI development in the UK and beyond. Artificial intelligence gave us machines that answer. Artificial societies will give us machines that act together. Managing that transition well is the defining task ahead.
Have a view on the matter? Share your expectations for artificial societies — and whether you would let an agent negotiate your next mortgage — in the comments below. To examine the research first-hand, order a prospectus or book a place at an upcoming open day and meet the academics shaping this emerging field.
