GUEST SUBMISSION: Remember when social media was going to connect the world? That was the actual pitch, somewhere back around 2006. Break down distance. Give everyone a voice. Bring us all closer together.

What we got, almost two decades later, was an app economy engineered to serve up one more Instagram Reel, one more outrage cycle, and one more reason to keep scrolling, because the algorithm learned that making you mad keeps you engaged.

One of Facebook’s own founding presidents later admitted that the platform was built, on purpose, to consume as much of your time and attention as possible. The same intermittent reward structure that makes slot machines effective was now sitting in your pocket. The infinite scroll has no stopping point because a stopping point is exactly where people would leave. Outrage gets amplified over good news because outrage keeps your thumb moving.

Attention was the product. We were the inventory, and “connection” was the marketing copy on the box.

That was not an accident. It was the business model. I am bringing it up at the top of a piece about construction technology because I think our industry is standing at a very similar fork in the road right now, and most of the conversation around AI completely misses it.

The demos are great, the delivery, less so

Every AI demo I have sat through in the last two years has the same shape. Big promise. Flashy stage lighting. A line like “this changes everything,” usually delivered by someone who will not be the one fixing it when it does not.

Apple has been promising a meaningfully smarter Siri since 2024, one that can actually understand personal context and act across other apps instead of just setting a timer with mixed success. That feature has taken longer than expected and has now been pushed further out than originally promised. Reporting has suggested that some of the new capabilities were failing in roughly a third of test cases during development.

Earlier this year, Apple also had to disable some AI-generated news summaries after the system produced false or misleading headlines. That means the assistant that could not reliably summarize a real news story is the same one that will eventually be trusted to read your personal documents and act on your behalf. I am sure that will go fine.

Microsoft is running a version of the same play, just dressed up in enterprise language. On investor calls, Microsoft likes to lead with the fact that roughly 70 per cent of the Fortune 500 has “adopted” Microsoft 365 Copilot. It is a great number, right up until you look at what it is actually measuring, which is largely access and licensing, not necessarily work being done.

A cleaner way to look at the number is this: Microsoft has disclosed millions of paid Copilot seats against a Microsoft 365 commercial user base in the hundreds of millions. That puts real paid penetration much closer to the low single digits than the headline adoption number would suggest.

Some studies also suggest that regular usage is still well below the licensing story. When employees are given a free choice between Copilot and ChatGPT for the same task, one recent survey found that roughly three out of four picked ChatGPT.

That is not an adoption story. That is a corporate gym membership. Purchased with conviction in January, forgotten by March, and still quietly showing up on the credit card statement in October.

None of this means the underlying technology is fake. It means the gap between the announcement and the actually delivered product is, right now, enormous. Most businesses are buying the announcement.

Our industry is not exactly leading the charge either

Before anyone in homebuilding gets too comfortable laughing at Apple and Microsoft, it is worth being honest about where construction sits on this same spectrum.

We love to talk about AI at conferences. We are considerably less enthusiastic about retiring the actual file named “FINAL_v3_USE_THIS_ONE.xlsx” that the entire operation has secretly been running on since 2012.

The British-based Royal Institution of Chartered Surveyors (RICS) recently put the industry reality in plain numbers: 45 per cent of construction organizations report no AI use at all, and another 34 per cent are still in early pilot phases. Less than one per cent have scaled AI organization-wide in any meaningful way.

When AI pilots fail across every industry, not just ours, they usually fail for the same boring, unglamorous reason. The data underneath was a mess long before AI ever showed up to read it.

One widely cited MIT report found that 95 per cent of enterprise GenAI initiatives showed no measurable profit and loss impact. Not necessarily because the models were useless, but because the tools were not integrated into actual workflows.

Garbage in, garbage out. Just faster, and with a nicer font.

The lesson worth stealing

Here is the actual point, and it is the reason I started this piece with Facebook instead of a spreadsheet.

Social media may have been sold as connection, but the business model rewarded attention. Over time, attention and value drifted very far apart. The product became engagement, not usefulness, and after enough years we barely noticed the substitution happening.

That is my concern with AI. Not that the technology is useless, because it clearly is not. My concern is that companies will use AI as another conduit for addiction and engagement, another layer of software designed to keep the subscription active and the user hooked, without ever delivering a meaningful lift in productivity or decision-making. 

AI is sitting at the same fork in the road, except this time every individual business gets to choose which version it buys.

That choice is not abstract. It shows up in very specific, very practical decisions inside a homebuilding operation.

You can adopt AI because it gets a subdivision agreement reviewed in minutes instead of an afternoon. Because it catches a margin problem on a live phase before it quietly becomes a write-down at closing. Because it gets a credible pro forma in front of a bank before your competitor’s team has finished their first cup of coffee.

Because it compresses the two hours you used to spend hunting through a 300-page planning report for one buried setback clause into a five-minute conversation.

You can also adopt it because the keynote speaker at the last industry conference told you to, because everyone else seems to be talking about it, and because nobody wants to be the builder who looks like they are behind.

That second version is not really adoption. It is FOMO with a procurement budget.

The test I would offer every builder reading this is simple. Ask what the tool actually produced. A decision made faster. A dollar saved. A risk caught two weeks earlier than it otherwise would have been. If you cannot answer that question in one sentence, you very likely bought the demo, not the product.

One of those paths is a power tool. The other is a slot machine with a nicer interface.

A power tool makes an already skilled person faster and sharper at the things they already know how to do. A slot machine keeps you engaged, keeps the subscription active and quietly avoids ever being asked what it actually delivered.

Use the tool. Skip the slot machine.

Maybe, while we are at it, rename that spreadsheet.