David England asks if AI can deliver on tasks in the workplace, and even more so, in health and safety.
Humans have been evolving tools to improve their lot for a quarter of a million years, invariably through a need to resolve some issue or other. Each individual development preceded by a variety of developments before it, and subsequently becoming the springboard for something even newer later down the line. The need to remove water from wells efficiently, originally done by hand, led to buckets being lifted by rope attached to a wheel driven by oxen, which was later improved by the Archimedes Screw which continuously rotated and was, therefore, more efficient still.
The wheel – the single most important technical development of mankind – was itself a progression from rolling heavy items on logs and led to the invention of cogs and, subsequently, gears. The Archimedes Screw led to the development of threads, which led to nuts and bolts. Imagine a world without wheels, cogs, nuts and bolts, if you can. Necessity has been the mother of invention throughout human history, with each technical progression building on those that came before it. Without the first flint knife, we would not have space rockets.
In professional worlds, too, there is progress through individuals testing and developing new ideas. In the world of safety we stand on the shoulders of Mayo, Haddon, Rasmussen, Hale, Reason among many others. Even your humble author of this article has postulated new theories on construction safety and assessing risk. Continuous improvement has been a rallying cry of human nature for almost as long as there have been humans.
But what of the currently ubiquitous artificial intelligence (AI)?
Credit: Steve Johnson/Unsplash
Its prevalence in all walks of life is hailed (by the enormously wealthy companies that market it, at least) as a hugely significant step. Its boast of accessing all knowledge to present practical solutions is edifying. But the ‘solution’ of AI belies the scramble to find new areas in which to use it. Is it, therefore, that significant? Or is it the answer nobody was looking for to a question no one was asking?
AI promises, by absorbing all of human endeavour through its large language models, to meet and exceed human knowledge. But if AI presents us with all the answers through what we have already learnt, who will drive knowledge forward from now on? Are we, in fact, drawing a line in the sand for what humans will think, invent, create, and dream of in the future?
Are we willing to leave the development of new ideas to the writers of the software that AI uses? Ask AI programs developed in China, the US, and Russia to each write a treatise on Tiananmen Square, Guantanamo Bay, and the invasion of Ukraine and see if they provide subjective analysis.
And what of AI in safety? Already there is software that can ‘analyse’ risks in the workplace and spew out risk assessments by the dozen. But is that all that safety is about? What of the control measures, the implementation, the training, the competency, the direction? And who accepts responsibility for software interpreting regulatory compliance? We saw during the CrowdStrike debacle earlier this year that software – and, moreover, software engineers – are not immune to colossal mistakes; mistakes for which, it would appear, there is no recourse to compensation.
Health and safety – proper health and safety – is a subtle blend of risk analysis, incisive questioning, appropriate knowledge, and psychology. It takes experience, competence, and passion – things which are rarely, if ever, found in electronic circuits.
