As the calendar flips over to December, it is the time to look back at the year that was and ahead to the coming year, guessing what will be the next big thing in 2025. We will have plenty of that at Geo Week in the coming weeks, to be sure, leaning on the expertise of industry thought leaders about what will be the talk of the geospatial and AEC sectors next year. That said, before even speaking to anyone it’s fair to expect that artificial intelligence will be a significant part of any look back and look ahead.
It’s hard to predict how we will look back on this era decades from now, but it’s certainly at least possible that we will look back at 2024 as the “Year of AI.” AI has been a prominent piece of the technology puzzle for decades, of course, and depending on who you talk to there is extreme growth coming in the near future, but this was the year when products really started hitting the market and it started to become a mainstay in many workplaces across sectors. As someone who covers technology, AI has been a topic in nearly every conversation I had this year.
Being in the time of reflection on the past and future, it’s gotten me thinking about where the path for AI is going in 2025, specifically for the AEC industry. In many ways, AI should be the ideal tool for this particular industry. Construction is dealing with oft-discussed worker shortages coinciding with increasing demand, leading to obvious productivity issues. It’s also an industry rife with massive amounts of data, much of which often goes unused.
All of these factors should point to AEC being the perfect industry to embrace AI, and if you talk to technology companies who serve the industry they’ll certainly agree. Every major player in the space – think Autodesk, Trimble, Procore, etc. – is adding more AI-enabled features within their broader offerings. It’s hard to imagine they would be doing that without some bit of assurance from end users that they are looking for these features.
I’m still wondering, though, if 2025 is the year we see a bit of a step back in the AI of it all in the AEC industry. That’s not to say that the technology will go away completely; the genie is already out of the bottle in this regard, and there’s too much value already being derived for the tool to go away completely. I do, however, think there’s a fair chance we see less enthusiasm around AI, and companies focusing their publicity efforts less on AI but rather zooming in more on specific capabilities of a tool.
With much of the fall being taken up by conferences around the globe, it’s a good opportunity to speak not only to representatives from the aforementioned companies providing these tools to the industry but those on the ground as end-user professionals in the sector. Thinking back on those conversations, there are a few themes that stand out giving me pause about where AI is going within the space.
One is something that applies to all industries concerning the use of AI tools, and that’s around the return on investment for all of the various products being released. There are massive amounts of money being invested into AI development and research right now, and there has been for a few years now. Some industries can give a longer tail on these investments to be sure, but construction is and always will be a relatively small margin sector. There isn’t as much time to wait for the ROI to show itself. This is already starting, and I think we will continue to see firms and other AEC organizations press for where the short-term ROI is going to come from, rather than just being sold based on theoretical value.
When you ask those who haven’t yet invested much at all in artificial intelligence to this point what’s holding them back, there’s a good chance that data security is still going to be at the top of that list. This, again, is a concern that exists for every industry, and AEC is always going to have issues here. This is an industry that is very secretive with its data given the intensely competitive nature of the firms fighting for a limited number of projects, the number of employees whose information is stored within this data on any given project, and the amount of government work that encompasses the industry. Data security is something that really has to be proven to be believed, and 2025 may be too early for that to really be proven.
There’s also the fact that when we think specifically about large-language models, hallucinations – when an LLM confidently relays incorrect information – are still a problem. This is something that came up during my attendance at Bluebeam Live, with multiple executives pointing out that they are still hesitant to adopt LLMs into their workflows because of their propensity for hallucinations. It’s true that, when working correctly, these tools can often provide real productivity gains, but any of those benefits take a swift step back if and when incorrect information is applied to a project.
One of the topics that has come up in multiple conversations with stakeholders when talking about AI usage within the AEC industry is the idea that the industry is scared of repeating its lag in adopting technology in the past. We all know the sector has that reputation, and some believe that leaders really don’t want that label to hit them again with AI, so they are jumping in early. That is, of course, not a good reason to invest in new tools.
As noted above, there are plenty of reasons why AEC should be investing heavily in AI. The benefits could be massive for the sector. That said, it seems like we’re moving toward a plateau in excitement and more movement toward practicality, which should ultimately benefit everyone. AI is here to stay and will continue to help an industry that needs to use every tool at its disposal to keep up with demand. Now that we’ve seen the first way of tools, the focus needs to be – and likely will be – less on the hype and more on the practical benefits.
