The sort of engagement that humans have been accustomed to MicroStation in recent decades is required by interactive software, which requires the user to be directed through a procedure. Many of these procedures are currently or can be automated as artificial intelligence advances. Machine learning can automate any non-automated step; how should a software architect determine the bounds of automation and acceptable system roles?

The user had to perform all of the work in the early days of ProjectWise interactive word processing-typing, layout, and checking. Spell and grammar checking are now regulated, and template-based automated layout is commonplace. What comes next? Will a natural language generator take human thoughts and facts and construct an appropriate narrative? This is currently true for weather predictions, and NLG technology is planned to give both a textual narrative and a commonly used "dashboard" of analytics tools.

With each interactive system, the user's role changes, and BIM Solution provider if the software being used is interactive, there must be an opportunity to integrate automation to boost user productivity. It is also vital to recall that machine learning enables automation without the requirement for task comprehension.

Is there no need to comprehend?

Consider traditional data entry, which involves transferring information from a handwritten form into a digital system. Unlike an optical scanner, a human would view the indistinct handwriting differently, squinting, stepping closer to the light, and maybe asking a coworker for feedback. More crucially, people recognize context.

If machine learning theory is used, a learning layer might be introduced. Scanning input photographs and then creating a link between those images and the output supplied by the data entry clerk would be required. When the correlation is satisfactory, change the learning layer to an automated layer to provide output containing all of the learnt correlations. Hey presto - automatic data input of all jobs from previously allocated user labor. Simply put, the most accurate model for automation is chosen through continuous model training and comparison.

That's the notion, and it's one that has many machine learning scholars talking. I've never heard of a comprehensive solution for automating machine learning manual data input activities, but this technique changes the software's language translation and raises the question of whether every user activity is required in interactive software.

Now is the time to be interactive.

In addition to the time-consuming numerical computations underpinning simulated structures or fluid flow simulations, engineering software offers fertile ground for interactive software - dynamic visuals, context-aware menus, quick-click reaction sequences to produce and alter design data.

Ansys exhibited their "Discovery Live" software this year. This makes advantage of the parallel processing capabilities of graphics processing units (GPUs) to offer simulation results in response to design modifications in an interactive manner.

This will alter the way product development teams operate.

Interactive simulation will integrate simulation into the design process, enhancing early design decisions and eliminating (or reducing) the need for lengthier cycles of follow-up simulation for a fraction of all available choices.

People must be properly positioned.

Smartphones are definitely ahead of the curve at this time, with many mobile apps discovering handy interaction sequences that make similar assumptions about user intent and eliminate the need for most options.

While significant, they are rather low-level considerations; specific design and optimization of options, as well as larger-picture considerations, are required.

Direct task automation (often implemented as new functionality); machine learning-based automation (possibly capturing and replicating existing ways of working); and converting existing "batch" software into interactive sequences (possibly enabled by special hardware, cloud capacity, or data access) are three future automation options.

To make these decisions, development teams must be able to observe how their interactive systems interact with other systems and business processes in their environment. The assumption that users will continue to perform the same things in the same way is an essential one to disrupt.