New BESS edition released

This article is the foreword by Peter Fritz AO to the new edition of BESS – the Journal of Behavioural Economics and Social Systems published by Global Access Partners. The cover artwork is a fragment of Peter Whitehead’s 2019 painting ‘Barrington Dreaming’ (Barrington Tops from the Allyn River trig station, New South Wales, Australia).

When I started my career in information technology back in 1964, an army of highly trained computer operators, programmers and systems analysts was required to tend machines so massive they occupied whole rooms or floors.

Looking at them with our 21^st century understanding, those early computers seem limited. And yet, the fastest ‘supercomputer’ at the time – the mighty CDC 6600 – could still perform an impressive 3 million operations per second. Computers were the province of governments, universities and major corporations. Crude as they now appear, those early behemoths heralded the start of a computing revolution.

Now we know, it was a slow start. We analysed manual systems, coded programs and processed data, fast. That was all.

Later, we set up databases and gave computers memory. But we still did no more than process data, only faster.

Over the last 60 years, computers have reshaped our lives as well as economies and societies, but their phenomenal data processing abilities remained a tool in human hands.

Then something new happened in 2023.

Artificial intelligence (AI), a development that slowly evolved in the background, suddenly birthed into the open.

Computers were no longer limited to faster processing and memory that never ‘forgot’ with total recall, but acquired a capacity to ‘think’. That is, much like in human thinking, computers were now able to recognise patterns.

The exponential growth and ubiquitous deployment of AI may herald another revolution in man’s relationship with machines, but this time as agents, and perhaps equals or more.

While traditional computers follow explicit, step-by-step instructions – algorithms – compiled by their programmers to perform specific tasks, AI systems use machine learning to recognise patterns and make decisions based on vast amounts of training data, often without explicit programming. The increasing complexity of these systems means they generate new and unexpected ‘emergent properties’, such as mastering tasks they have not been exposed to.

AI can interpret data, find hidden patterns, make predictions and potentially offer novel solutions to problems we could not anticipate. Furthermore, while present-day large language models may still have theoretical limits we can only incrementally improve, future AI systems may undergo ‘phase transitions’, in which their abilities increase unpredictably and exponentially in realms which remain unseen.

However, AI models have no concept of human empathy. Aligning their ultimate goals with human needs may prove a more difficult philosophical task than a software engineering problem. AI therefore introduces an important new factor at the nexus of behavioural economics and social sciences, as well as ethical concerns regarding bias, privacy and autonomy.

AI machines, much like humans, make mistakes. Hence for AI to be a tool to benefit society, humans will still need to have subject matter expertise. For humans to be up to the task, our education system will need to adapt to the new reality and embrace critical thinking.

As we cede ever more of our cognitive labour to machines, will we expand our imaginative horizons or let go of the very faculties that freed us from our caves? The ability to reason, to create, to empathise – these quintessentially human traits may wither or blossom on the vine of artificial augmentation.

For good or ill, this technology will not be constrained, as theoretical inquiry, commercial pressure and geopolitical imperatives will continue to accelerate its abilities beyond the grasp of policymakers, and perhaps its designers too. Creating AI in our image could see it inherit our biases, our cruelties, and our capacity for self-destruction, but giving birth to what amounts to a new form of life might have greater ramifications still.

It is therefore even more important that we discuss AI’s ramifications in personal and social terms, as well as marvel at the technology or harness it for commercial gain.

BESS can become an important platform for these discussions, and I look forward to your views and contributions.

BESS Volume 6, Number 1 is available on Aalborg University’s Open Journal System and Global Access Partners’ website.