Responsible Use of Artificial Intelligence
- Duncan Cartlidge
- Sep 23
- 2 min read
Responsible Use of Artificial Intelligence
Generated artificial intelligence is not new to construction. In the 1990s expert systems were developed by the RICS and the University of Salford. They worked on the basis that human experts recorded their solutions to a set of scenarios, such as which procurement path to use, into an inference engine.
The clammer to jump on the generative AI band wagon reminds me of the rush to join the e-commence world at the turn on the millennium. In the 1990’s the must have toy for consultants and contractors was a website together with an online presence or be left behind. Nevertheless according to Building, that’s the same Building that has predicted the demise of the QS on several occasions, 84% of the top 150 consultants intend to invest in AI in the future with digital optimism at an all-time high. Currently the technical press is full of articles with pictures of rather smug looking individuals telling readers that they are already too late although many of the organisations who are enthusiastically trying to sell AI use words like; potential, expected and likely. Now, don’t ask me how anyone could possibly know this, but the ONS has produced a spreadsheet listing the UK regions most likely to be impacted by AI with for example, 8363 jobs being under threat in Stoke-on-Trent. In addition, the ONS has predicted that quantity surveyors have a 0.273715% chance of being automated, with construction project managers having a slightly better 0.257037 chance – I’ll take that.
It is reported by some early adopters of AI that they are enjoying a cost reduction on manual estimating. Now I remember the call for qs’s be confined to history when it was stated that BIM could produce bills of quantities however, after a few attempts it became evident that BIM generated bills needed a qs to check the output. Would you let an AI generated estimate leave your office without human oversight?
The RICS issued a global professional statement this month on Responsible Use of Artificial Intelligence in Surveying Practice in which it warns;
AI system outputs may not always be fully predictable or fully explainable, and in any event, there is an inherent risk of error and/or bias in AI systems due to error and/or bias in the algorithms used. This could be due to variable training, data quality, biased programming, lack of data diversity and/or the limitations of the AI system itself, as well as due to the AI system responding to or interacting with any biases of the firm in the context of its use.
Generative AI is only as good as the data it is fed on, just like the early expert systems. What’s the saying? - 'Rubbish in – Rubbish out'. So, if you’re scratching your head about just how to take advantage of the AI revolution, assuming it arrives, the smart people are letting the early adopters make the mistakes whilst getting their data into a form that any AI based process or system that they may wish to adopt in the future can fed on it to its heart’s content.
Duncan Cartlidge
Comments