Challenges and potentials

... of Artificial Intelligence in the Construction Industry - Results of the Roundtable on 19.06.2020

On June 19, 2020, the first digital roundtable of SDaC took place under the title "Artificial Intelligence for the Construction Industry - Potentials and Challenges" within the framework of the EU Digital Day 2020 (www.digitaltag.eu). The event was organized by CyberForum e.V. and moderated by Dominik Steuer (executive officer of SteuerTiefbau GmbH) and Svenja Oprach (project manager of SDaC). 

A total of 24 people from the construction industry (75% of participants) and other ecosystems (25% of participants) took part. 

In short surveys a common picture of the challenges and potentials of AI for the construction industry was compiled. The results are presented below (multiple selection was possible):

In your opinion, what are the biggest obstacles to successful digitisation of the construction industry? (n = 21)

As the three biggest challenges, the participants named unclear interfaces (76%), insufficient IT knowledge among technical personnel (67%) and the quality of the digitalized data (62%). For successful digitization, it is therefore necessary to prepare data, transfer information across system boundaries without loss, and provide intuitive applications for everyone. 

This confirms the concept of SDaC: Heterogeneous data is prepared and linked via interfaces on the SDaC platform. On this basis, human-centered, intuitive applications can be designed. 

The SDaC concept from heterogeneous data to intuitive applications

What are your expectations when implementing AI in the construction industry? (n = 21)

According to the participants, especially repetitive processes can be supported (95%). As an example, the use case "Object recognition in building planning" can be mentioned here: If a planner or site manager searches for individual elements of the same type in plans, a trained AI can reproduce a first list of all elements in a fraction of the time and in real time. As a result, creative activities can be expanded (62%) and complexity (67%) reduced. The intensification of customer relations, like those between a construction company and building owners, was not primarily seen as an expectation. Furthermore, the following expectations were mentioned in an open discussion: The reduction of errors and the establishment of new business models.

What risks do you see in implementing AI in the construction industry? (n = 20)

The parties considered the market dominance of large construction companies due to the availability of large amounts of data (70 %) to be a risk. In order to develop the construction industry as a whole in a sustainable manner within the framework of an AI strategy, a special focus must be placed on SMEs. In Germany 99% of the companies are SMEs. New research approaches, such as machine transfer learning, can provide a way to exchange metadata via a platform without releasing the raw data. 

Potential misinterpretations (60%) were seen as another major risk. Unfounded and incorrect decisions based on an AI application can result in high follow-up costs, conflicts and quality deficits.

In this case, is the construction company as user, the data supplier or the provider of the AI application guilty? By means of explainable AI methods (white box) the SDaC user should be involved in the decision process, because he is the only one who can make the final decision and is therefore responsible for it. 

Other risks include the "war of data" (who has the better data?) and the need to build up know-how in the construction industry.

Due to the positive feedback, further digital roundtables are being planned. 

Take part in our next Roundtable on September 18th from 10:30-12:00 o'clock The topic will be the successful collaboration of humans and machines. You can register under the following link: https://register.gotowebinar.com/register/2076799752368934671