AI in Construction Workshop

Constructing Excellence

On Wednesday 19th July, we were delighted to be joined by Unit 9, Glideology and BRE to explore AI in Construction and the potential applications.

To begin the session, attendees were asked what AI means to them and how comfortable they are with the idea of an AI future. The results indicated that while people can see the potential benefits of adopting an AI approach, some mistrust of the technology remains.

 

Throughout the session, the group gained a better understanding of how AI can be applied and how we can move forward in building trust in the technology.

About AI

What is AI?

Artificial intelligence is best seen as ‘a collection of techniques that have been applied to data to try to extract models of the real world and shortcuts with minimal human intervention’.

 

Different Types of AI

  • Programmable AI (expert systems, natural language processing, vision, etc.)
  • Machine Learning (ML): statistical techniques to give computers the ability to learn with data. It can further be categorized into supervised learning, unsupervised learning, semi-supervised learning and reinforcement learning.
  • Specialised new areas that combine techniques, such as LLMs, GANs and computer vision.

 

 

Why Are the Special Areas of AI Important?

Challenges Within AI

  • Production problems:
      • Unexplainable- lack of transparency around AI’s decision making/reasoning process.
      • Bias- this can be reflected in AI systems if imperfect training data is used.
      • Data privacy- need to protect personal data as AI will collect more human data for improved training.
      • Data dependence- relying on large data sets which may not represent real world.
      • Regulation- determining regulations and standards for ethical AI use.
      • Safety- creating AI that is safe, secure, and aligned with human ethics and values to prevent harm.
      • Technical complexity- such complex technology can be hard for even the experts to understand.
  • Social problems:
      • Job displacement- AI taking over for human workers.
      • Limits of AI- recognising AI limitations in emulating humas- not anthropomorphising.
  • Civilisation factors:
      • AI ‘pseudo-intelligence’- overestimating AI capabilities and challenges to achieving general intelligence.
      • High resource needs: need for extensive computing power, data storage and energy for complex AI development and deployment.

These challenges highlight the work that needs to be done around AI for the construction industry to embrace the technology and begin to employ AI practices. Without data security, resources and recognition of the limitations, AI cannot be implemented effectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

In your sector, how much do you trust AI?

Glideology

TRAMS-CONSTRUCT Project

Glideology have been leading on a 3-month feasibility study, alongside other SMEs and BRE, about the use of AI in a trustworthy manor. This project is called Trams-Construct and aims to bridge the gap between AI and Construction with an objective to create a platform that collects and pre-processes on-site data from construction contractors managing large sites. It focuses on enabling Trustworthy and Responsible Artificial Intelligence (AI) and Machine Learning (ML) technologies to address challenges faced by construction contractors on large sites.

 

Trams-Construct aims to tackles the challenge of data security within AI and enable more trustworthy application of the technology. The findings from the work will be compiled into a White Paper and, if successful, will move onto phase 2 with funding from Innovate UK.

Discussion

How can we build trust?

Trustworthy AI needs to allow for human intervention, e.g a Super User who can step in and make an overarching decision if the AI’s suggestion is deemed unsuitable. This puts responsibility with the decision maker, who can provide rationale for any actions taken- AI is unable to do that. However, AI can offer multiple options which humans may not be able to see, which would support the decision-maker.

Enabling ultimate decision making for the human will support the building of trust over time. A history of accuracy will build, proving that AI can and does get it right and enabling humans to rely more on such solutions.

 

Where Can AI Make the Biggest Impact?

By tackling problems that the industry already has. The next generation of AI will be interesting as it’ll begin to identify and solve problems that we, as humans, were not even aware of. This means it will be able to prevent problem before they actually occur.

 

How Do We Get People in Industry to Embrace AI?

Generally, people welcome things that make their life easier. In life we hand over a certain level of decision making to technology, i.e. google maps or Sat Nav, this new technology will eventually be used in a similar manor.

 

Click here if you would like to view the workshop slides.