How civil engineering can use AI to step up its digital transformation

When you hear the term artificial intelligence (AI), what emotion does it provoke?

Excitement? Apprehension?

The public release of ChatGPT in November 2022 sparked both fear and wonder among the engineering community towards AI’s wider adoption.

It’s no secret that the amount of research and powerful potential of AI continues to increase year on year.

And with it, the number of commercially viable civil engineering uses.

Regardless of how we feel towards AI, if we begin by demystifying what it’s all about, we can all step up our digital transformation.

How is AI applied in civil engineering currently?

AI already creates opportunities to significantly reduce computational processing. This is the effort required to complete calculations usually measured as unit power consumption or time needed to calculate with a model of computer chip.

AI does this through surrogate modelling.

This is when the model is trained with known input/output combinations, so that the AI can predict the results and infer the development of leading indicators from lagging data.

This can benefit design through improved soil behaviour understanding (i.e., liquefaction potential, in-situ soil properties) and operations (i.e. big data flood detection and forecasting) to name a couple.

In many cases, this can decrease computational time by more than 98%.

Below are some more of the already established uses of AI in civil engineering.

AI is only as good as its data

Rapid prototyping provides civil engineers with more opportunities to test pioneering sustainable solutions and optimise designs.

Better operational predictions can improve asset management for the overall system within built and natural environments, while having the potential to save lives, as is the case of flood prediction.

But the AI will only be as good as the data it is trained on.

For new, complex, or extreme conditions, it will interpolate the result.

The greater the difference, the worse the overall reliability of the output.

Therefore, in this case, the AI models should be suggestive, rather than prescriptive, when making decisions.

How is AI used in construction?

For engineering contractors, AI is enabling opportunities for modern methods of construction (MMC) to improve delivery efficiency and safety.

To name a few developing uses:

• identifying safety hazards from site CCTV
• reactive construction scheduling
• off-site construction component assembly simulation
• remote inspection using virtual reality (VR)
• optimised labour management
• cost overrun predictions

All in all, AI is reducing human error, optimising spend, creating a safer environment, and setting up necessary conditions for further predictive indicators to be developed.

Thus, it’s increasing the overall productivity of the construction industry through accelerating digital transformation.

Where does ChatGPT fit in the AI space?

If civil engineering is already acquainted with AI, why is ChatGPT felt to be so significant to so many?

ChatGPT, built on GPT-3 technology, is a large language model (LLM), flourishing in its contextual language competencies.

This enables it to learn from text data (over numerical or visual) and provide human-like responses with immediate value to the user.

Rather than traditional AI models being only useful to the one task they have been trained on, GPT-3 is proficient (to a degree) in any task involving text.

To better elaborate on it being proficient in any task, the table below provides some insights into uses identified by engineers via LinkedIn and discussion forums.

All of these have been popularised, trialled, and are being worked towards proactively and publicly since November 2022.

Establishing rules in order to step up our digital game

The uses of ChatGPT, and AI in general, overlap with nearly all professions and activities in the civil engineering industry.

This removes doubt that civil engineering will see anything but a continually increasing rate of adoption and proficiency with AI assisted deliveries.

‘The rules of the game’ therefore need to be proactively updated by the asset owners, regulators and governing bodies.

This would enable an understanding of their position towards AI, and an appropriate balance between optimisations and new risks it can bring to projects.

Without such interference, there’s a risk of civil engineering expertise being diluted in a race to the bottom market situation – where AI improvements compound time savings with automated checking and standard delivery templates.

This would force consultancies to aggressively reduce contract fees as other providers can do it for less using AI.

Considering this risk, regardless of whether you’re afraid or excited for AI’s wide adoption throughout the industry, it’s time to step up our game.