Read the post as published on Construction Computing.
ALICE Technologies teams up with Oracle as AI continues to transform construction with the introduction of ALICE Core.
It’s not the answers you should be looking for – it's the questions. This was explained to me by ALICE Technologies CEO, René Morkos, when I spoke to him recently. Construction schedules are incredibly complex, involving many variables, different technologies, alternative materials, fluctuating costs and fluid resources. Charting the most efficient and cost-effective route for a construction project is a skilled job, supported by well-established scheduling software such as Oracle’s Primavera.
It’s an ideal task for Artificial Intelligence, which can analyse thousands of possible permutations, balancing the requirements of each element of the construction project to arrive at an optimum solution based on different criteria – cost, resource availability, construction sequencing, delivery constraints, holidays and weather limitations – in fact, any variable that you can throw at it that would affect the schedule. A straightforward task for AI as long as you can tell it what you want it to do.
René described what you need to do before you can use AI to optimise a construction schedule. “First of all,” he said, “You have the planning phase, where you have to take the drawing or 3D model, identify the scope, and figure out what you are building – and then work out how you are going to build it by converting the scope into a list of tasks.” AI won’t figure out what it is you are trying to build, he explained, or whether you would use precast instead of cast-in-place concrete, or what production rate you can achieve.
Scheduling assigns start times and resources to each of those tasks so that you don’t violate constraints. Because of this, planning and creating a single schedule for one construction project is quite a major task – ideally suited for artificial intelligence, though, that can compare tasks and schedules, simulating the construction to arrive at an optimisation that satisfies the constraints that you have established.
ALICE Technologies Core
ALICE Technologies has been working on integrating AI technology into construction projects since 2015, when they introduced ALICE Pro, which enabled users to connect 3D models with their schedule and estimates to visualise all aspects of a project in 4D.
Recently, they made significant changes to their software, introducing ALICE Core, which takes AI mainstream, enabling it to directly import P6 and Microsoft Project schedules into the tool. It’s now fast and easy to bring information from Oracle’s and Microsoft’s scheduling products directly into ALICE and back again.” René said.
ALICE uses AI to analyse a project’s complex building requirements, generate highly efficient building schedules, and adjust those schedules as needed during construction, simulating thousands of options in seconds, and testing different scenarios to find the optimum solution. It leads to huge time and cost savings and reduces risk on projects.
And there’s the crunch. Allocating tasks, resources and start times won’t differ much whether you use AI or not. Changing any one of these elements manually, however, would involve reading, redrawing and re-creating the links between each variable – for 6,000 tasks, or so for a straightforward multi-storey project. With ALICE, all of that would be done automatically, and René quoted a typical afternoon’s work where 600 million different schedules can be run for half a dozen to be selected for implementation.
The real benefits start to occur in real-time, when any one of many scenarios can occur – such as building materials not being delivered, manpower shortages, whether you can reduce costs by shortening the schedule, and other trade-offs, each of which can be explored instantly.
ALICE Pro versus ALICE Core
I asked René what the difference was between ALICE Pro and Core, and how easy it was to get up and running with AI software. He explained that ALICE Pro had required a week or so of training, learning to use it like a new language. ALICE Core, however, is far more intuitive. René said that within half an hour of downloading ALICE Core, and with a minimum of set up or training, new users can bring in a P6 or Microsoft Project schedule and start to use it.
Partnership with Oracle
With the launch of ALICE Core, ALICE Technologies has also joined Oracle Partner Network (OPN). This allows construction professionals using Oracle’s Primavera Cloud, Primavera P6 Professional, and Primavera P6 EPPM scheduling products to optioneer their schedules using ALICE.
As a result, customers can bring enterprise workloads to the cloud quickly and efficiently while addressing the strictest regulatory compliance requirements.
“The introduction of ALICE Core provides additional value to construction professionals who rely on Oracle’s established scheduling solutions,” said Frank Malangone, Executive Director, Industry Strategy and Innovation, Oracle Construction and Engineering. “We’re pleased to be collaborating with ALICE to help our customers leverage an AI-based tool to optimise their construction schedules, and we welcome ALICE to Oracle Partner Network.”
ALICE Core will also be made available with other scheduling applications over time. Prospective schedules don’t involve very complicated file formats – they're basically Excel spreadsheets with lists of tasks, start times, durations and relationships, and, as such, could be imported to ALICE as CSV files.
How will AI develop?
AI is constantly evolving. According to René, who has been working with it for the last 20 years or so, it is a convenient term to describe the latest set of algorithms that have been powering advanced calculations for many years – the term was originally coined in 1954 at Stanford University. It has been attached to constraint programming, discrete event simulation, machine learning, and, now, neural networks, but they all describe combinations of algorithms to solve different problems – considerably smarter than they used to be, but still algorithms. According to Moore’s Law, which René quoted, the power of computers will double every 18 months, which means that the 2K computers I was working on in 1970 will be billions of times more powerful now – and, as they and associated technologies become more efficient, faster and powerful, we shall see neural networks, etc., overlaying old school algorithms to perform even more complex tasks.
The problem is, as René says, ‘they haven’t got a bloody clue what they’re doing unless they’ve been told to do it by the humans who provided the algorithms’. Which leads us back to the opening sentence. ‘How do you get AI to start making the decisions?’ I read somewhere recently which confirmed René’s position. ‘The answers in the book are subjective and therefore unimportant, go seek the questions, instead’.