The remit of artificial intelligence (AI) is to develop machines that have potential beyond the ability of humans. Its primary focus is on productivity, forecasting, data management, optimisation, automation, picture interpretation, speech and face recognition and dialogue.
AI has been challenged for errors that compromise safety, level of transparency, intellectual rights, discriminatory face recognition as it is predominantly trained by white people and data protection. Most cases are adjudicated via litigation and, as AI legislation is still at a premature stage, the jury is still out. But when it comes to ethics, there are more parameters for consideration than merely accountability allocation.
Building information modelling (BIM)
The ability of one person to control all information at any stage of a project is often overestimated. Not only because of the constantly changing and complex nature of the construction world, but also because of the numerous reporting streams. The ongoing development of BIM to act as a single source of truth along with automated reporting tools gave some solace to project managers.
Design managers have also realised benefits through the use of 3D federated models that enable clash detections and multidisciplinary designs to be signed off with higher confidence. Construction managers in Dubai, the Netherlands, Spain and China are starting to benefit from a far greater relief, that of 3D printing directly from the 3D model. Increasingly we see people departing from the old-fashioned “proprietary-rights” mindset silos and working more collaboratively to achieve a shared vision. The efficiencies achieved through the implementation of AI give engineering managers more time to invest in modernising governance, workplaces and designs.
Construction AI also has potential in the commercial aspects of a project that stretches beyond its current ability to predict cost overruns and programme delays. In an ideal world, data will automatically be pulled from databases of material procurement, resource utilisation and drone footage and all transactions will be progressed digitally. An innovation that leaves very few chances for budget mismanagement whilst enabling cost reduction and transparency.
Technological unemployment or human liberation from hardship?
In a simple example of brick laying being carried out by robots, benefits are realised from eliminating what the HSE recognises as the highest rated risks; falls from height, people and plant interface and occupational wellbeing. Getting it right at first and achieving defect-free end products within a shorter timeframe would be part of the picture. There will be no need for workers, operators or surveyors as these machines would be an amalgamation of their skills. On the unpredictable day that these robots will become both consumers and prosumers of energy, construction will be proud to be heading towards zero marginal cost operations.
Regardless of these advantages, the public’s concern on job security and lack of understanding of AI could potentially label AI as “immoral” and render it unwelcome.
AI triggers multiple novel needs that create job opportunities, some examples being within auditing, longer design, coordination and planning periods, material innovation, policy development, supervision, trialing and green energy.
Thereafter, corporations are called to place corporate social responsibility at the heart of their strategic plans, act proactively and view themselves as part of the solution, rather than part of any skills-shortage problem. Unless the pace at which people gain alternative skills outpaces the rate of skill replacement, we are set to prove that the detractors of AI are right.
How much accountability can we allocate to AI?
When ethics is taught, one of the most common case studies is where you are asked to visualise that you are in charge of a lever that controls the direction of a train. The train is heading towards a group of people. If you pull the lever, there is one person standing in the alternative track. There is no escape route in either case. What would you do? Would you sacrifice the one at the expense of many or do you refrain from taking any action? How would your decision change if that single person was a family member or an indispensable asset to the profession?
Are we expected to pre-empt prejudice and feelings when making decisions or are moral decisions a product of emotional intelligence? The responses on what the right thing to do is span across three principles; utilitarian, whereby decisions are taken in the merits of net benefit, the action theory that supports inaction to occurrences and finally the intentional principle which delves into the deeper reasons for our choices.
If AI was set to decide on this dilemma, this is no suggestion that humans are no longer responsible. Fundamental decisions such as what data should be instructed to the machine, what models and algorithms should the machine be applying at different scenarios and how will data be collected without any infringement are still under the power of humans.
Technology has no morals
AI is a human construct and therefore subject to human error. Unexpected events, data misinterpretation or mismanagement may trigger failure. A small error within highly risky projects may trigger irreversible damage.
I recall the story of the self-driving car where on a bright day, the sun rays’ reflecting on a white van confused the car to assume a free lane causing a collision. The use of AI requires surveillance and education on its limitations, ongoing betterment of interpretation and prediction, and systematic error autocorrection.
Advance agreements between all parties with context similar to the responsibility assignment matrix will reduce the number of cases at the court and increase investment and interest.
Is AI a decision of the amoral?
Whether further deployment of AI enhances ethics, is a moral decision that should be taken by both the public and technologists. To answer this question, we need to ask other questions: What are the benefits and pitfalls? Who benefits from technological evolution? Do benefits address global inequalities? Is AI the most sustainable option? What would be the optimum level of introduction? If the global benefit outweighs residual risks, then it is the right way forward.
- Ioli Kynigou is a senior manager with Systra and is also a member of the ICE’s Ethics Committee
Like what you’ve read? To receive New Civil Engineer’s daily and weekly newsletters click here.