Come Tomorrow, Part 2
Vince Bailey / June 2019
It’s the end of the world as we know it.—REM
Estimators, you should stop casually paging through this issue and zero right in on the following prediction: Your function as we now know it will be obsolete inside of 10 years! Wake you up? If you’re skeptical, as I expect you are, cast your memories back a bit. Consider the developments we have experienced in the complexity of our estimating programs over the past 20 years. Nothing less than staggering, eh? Now apply that eye-popping development forward, factoring exponentially for building upon the strides already made. If you’re even slightly imaginative, the future possibilities over the next decade will take your breath away. And the engine behind this whiplash thrust into Tomorrowland is already being assembled. It all rests upon the ongoing development of artificial intelligence.
Last month we explored the pros and cons of AI and drew a fuzzy line to the conclusion that it will be the death knell for the estimating profession as we know it. This month I will attempt to brighten up that fuzzy line and make a direct connection. But first, I think a brief recap is in order.
We should start with a basic definition of AI: artificial intelligence is the simulation of human intelligence by computer systems—i.e., systems so advanced, they can mimic human cognitive processes and are therefore capable of decision-making, reasoning and even self-teaching. AI will be able to catalog, analyze and manipulate huge volumes of data quickly, tirelessly (works 24/7), effectively and error-free, thus freeing humans to focus on the creative side of processing, which happens to be AI’s weak spot. Now, optimistic readers will interpret all of this as a boon to mankind, freeing us all up from the drudgery of number-crunching. But when number-crunching is what you do for a living, the looming advance of AI may be harbinger of doom. And while those of a more sanguine persuasion will point to the certain creation of whole new professions, the estimated net loss of jobs due to AI over the next five to 10 years is 400,000!
Contemplate, if you will, what we estimators do to justify our existence. We measure and quantify by standardized units (lengths, areas, counts) certain constructible assemblies through symbolic designations (lines, colors, spots and field sizes). We take these accumulations, add or delete certain components to conform with a designer’s criteria for the assembly, and assign them a value with regard to material and estimated labor to provide and install such an assembly on a given project. We tally up the totals, write up a summary regarding scope and terms, and provide it all with a markup for our time and trouble. Then, voila! That’s pretty much it in a nutshell, isn’t it?
Now consider a hypothetical: What conditions would an AI program require to perform the same functions we bidmeisters do based on the above job description? Suppose our industry agreed on certain standardized assemblies for estimating (or quantifying) purposes. Suppose we all subscribed through a common program to a common super-database that comprised templates for every possible assembly with every combination of components possible. Most of our current programs utilize templates for the most commonly used assemblies (let’s say wall types, for example), and we presently cherry-pick from thousands of components contained in a large database to custom “design” our assembly per conformance requirements by adding or deleting certain line item components. But what if that step could be eliminated by simply typing in a code that represents that unique combination of components—a pre-combined assembly from our hypothetical super-database? Of course, such a massive database is currently inconceivable. But then, the development of AI, by its very definition, claims the ability to manipulate vast amounts of data. Good thing, then, for us estimators, that the cost of such a database is well beyond the means of even the biggest drywall subs.
Who, then, has the resources and the inclination to be the first to incorporate AI into their endeavors? Well, in the world of preconstruction, architects are at the top of the food chain. The million-dollar price tag for a custom-designed system is no obstacle for many top design groups. And the lure of a program that will free them up from the drudgery of converting aesthetic concepts into constructible projects will have an irresistible appeal to their vanity. After all, it is quite unfair that their creative minds should be regularly wasted on repetitive and mundane tasks. Already, several top design groups are utilizing AI within their own scope of work. How long, though, before they develop that super database described above—one that codes and selects wall types and other assemblies based on performance criteria, one that integrates selections from other databases that track current material prices and labor productivities? Beginning to feel vulnerable? Author John Pugliano, from his 2017 book, “The Robots Are Coming,” put it this way: “Bottom line, any routine job that can easily be defined by a mathematical or logical equation will be at risk.”
Next month, I will describe in depth an integrated AI system, like the one above, that could reduce the estimating profession to virtual obsolescence, and will radically change current protocols of project delivery.
Vince Bailey is an estimator/project manager working in the Phoenix area.