In December of 2022, ChatGPT seemingly came out of nowhere to grab everyone’s attention, and soon, DALL-E demonstrated that generative AI was more than ‘just’ words—it could take words and create original images. Generative AI was the story of 2023. 

Of course, these generative AI miracles didn’t come out of nowhere. They developed out of a convergence of the necessary developments in natural language processing, machine learning, computer vision and giant network transformer models, all of which rely on large-scale computational capability. 

The computing infrastructure has finally evolved to the point that it can support the necessary processing capability required by the components of AI. We now have systems that meet the intensive computing requirements to build, run and maintain generative AI algorithms. Meanwhile, the internet en total is a prodigious resource for information input. ChatGPT trained on the public internet along with other sources. 

All those elements together allowed the transformation of chatbots perched on learning models into the generative tools ChatGPT, DALL-E and others. 

Leaders scrambled. Should they fear it? Leverage it? (how?) Shut it down? Allow some experimentation? 

“Yes.” That was the answer from Gartner Global Chief of Research Chris Howard. "The truth is ... all those things at once." This generative AI "doesn't know what it means," Howard says. "It knows how to speak, but it doesn't actually understand what it is saying." Therefore, guardrails are necessary. At the same time, experimentation is necessary. As has been demonstrated in these pages, generative AI can be harnessed now to facilitate otherwise tedious "shovel work" in a variety of ways. Individuals across all kinds of disciplines should and can be exploring applications of the technology, while understanding its limitations.

It's what adventurers in the technologically reluctant construction industry have done and are doing, as we have reported: finding ways to harness the available technology, being aware of its limitations, and increasing productivity in specific ways, from creating safety briefs to running project risk analysis.

Construction AI is here to cover it, bringing you the latest developments and applications together with step-by-step examples of how construction can harness generative AI in large and small tedious tasks to take productivity to new levels. 

A generative AI tool like ChatGPT is potent. Its blazing speed is fantastic. But despite the (genuine) wow factor, you must not assume that the output it delivers in literally seconds is complete and accurate. Human expertise is very much still required. That's the good news, at this point, for those worrying about being replaced. The trick is to see it as the tool that it is and learn to use it.

At the start of 2024, where do things stand? According to Gartner’s Hype Cycle, generative AI is  past the “peak of inflated expectations” and is already in the “trough of disillusionment.” And, yes, that was quick. Gartner’s Howard says, “This technology is moving through the hype cycle faster than any technology that I can recall.” 

The trough of disillusionment is the real beginning. It’s where the technology has collided with expectations and reality, but also where people wrestle the new technology into actual value-adding applications. The "trough," then, is “where the hard work starts,” as Howard says. 

Generative AI will take its place among other technologies, including AI technologies like machine learning that have been around for a while. Generative AI is in the category of artificial narrow intelligence—more on that later—and is set to become a very useful tool to support getting things done. Stay tuned.