Monday
May142018

Leading Thoughts: 5/14/18 Could Geotechnical Engineers be Replaced by AI?

Charles Allgood, P.E., LEEP AP considers AI and the future of geotechnical engineering.  This will impact Geopier ground improvement and every aspect of underground construction.

It seems that everywhere you turn there are conversations, articles and pundits discussing Artificial Intelligence (AI).  Some point to a bright future for humanity while others are pessimistic and talk about AI replacing people and doing their jobs better.

Young adults face an interesting dilemma that those of us currently working did not have to consider: choosing a career that will not soon be automated. The McKinsey Global Institute estimates “that half of today’s work activities could be automated by 2055.” This made me wonder about AI and automation in the geotechnical engineering profession. Who might be replaced and what would the workplace look like years from now?

While we explore this topic I am also going to ask you the reader some questions.  It would be interesting to get your perspectives.  There is a link at the end of the article to send your ideas and I will write a follow up article highlighting them!

There are geotechnical data gathering/testing methods that might be automated; we’ll explore a few possibilities.  We also know that geotechnical engineering requires judgment; this is based on experience, knowledge of the local geology and contracting methods, historical knowledge, and other variables.  But could the “human” element be captured by a self-learning AI under the tutelage of those experienced engineers? What do you think?

Here are three tasks in geotechnical exploration that could be automated relatively soon; afterwards we will talk about areas that may be difficult to automate.

Three tasks that may be automated:

 

Data Collection

Geotechnical engineers collect information on soil, groundwater and rock conditions.  Perhaps going forward the exploration begins with an AI searching for the available digitally-stored open-source data on historical property usage, the boring and well logs stored on agency or firm computers, aerial and old photographs/survey data, etc.  This information would help characterize the site before the exploration begins.  There is a lot of information out there and AI should be able to find it much better than we could. How many times have you drilled borings on a site only to find out that others have drilled there before you?!

Afterwards the field work could be supervised remotely by AI with robotic drill/sounding rigs performing the work. Do you think this is possible soon? If un-manned rovers on Mars can determine chemical and composition characteristics of soil and rock how far away are we from doing much more here? Drilling is a dangerous business so automation would be attractive. 

Soil Classification/Lab Testing

The field sample containers could be brought into an automated lab by aerial drone, with bar codes to identify the individual samples.  It should be possible to develop robots with the ability to test soils for texture/grain size, moisture content and even plasticity.  An AI could then assign additional lab testing based on project needs.

The structure for this is already in place with cross-over testing and quality inspection technologies from other industries, automated soil testing machines, advanced robotics and electronic boring logs.

Are there other tasks you think will be easy to automate?

Geotechnical Report Writing

Using the collected data AI could complete the analyses and engineering report.  This could be similar to the AI tool Reuters is building.  The software is already in place to analyze bearing capacity, settlement, stability, overlapping foundation stresses, earth pressures and other more complex 3-D modeling.  The AI would already “know” the entire library of geotechnical knowledge and have that to draw upon.

The AI could select foundation, floor slab, pavement, slope, retaining wall and other structure support recommendations.  These would be based on the initial client needs for settlement control, floor deflection, LEED® certification, or any number of requirements.

During the AI learning process human engineers would evaluate whether the recommendations are practical and advise on how to proceed.  Every time thereafter the AI should learn and get better.  Hopefully they won’t get too good at this!

Think about how long it takes to train a young engineer out of school to become a productive, confident engineer.  It is a long process.  Sometimes it doesn’t happen.

 

Two tasks that may keep humans in the game:

 

Quantifying Soil and Groundwater Conditions

We geotechs have an advantage; we work with heterogeneous materials.  Soil and groundwater conditions change, sometimes abruptly, between the sampling locations.  Also, when you add mixed unknown fill to the equation things become even more complicated.

This variability makes it difficult to characterize a site.  A good geotechnical engineer can use judgment to help “bridge” these data gaps.  Perhaps AI would use statistical algorithms as their “judgment.”

Perhaps Dr. Karl Terzaghi’s observational method will help keep us humans at the helm, or at least heavily involved.  When construction starts we will be needed to confirm the soil/groundwater conditions match up with the exploration results.  The initial site characterization and exploration won’t find everything. We can partner with the AI to evaluate whether changes are required to the recommendations.

It would likely be difficult to have machines perform these site observations.  It would require a blend of sophisticated robotics and programming.  I bet we will still be needed.  Do you agree?

 

Process Supervision

Geotechnical engineering can be a bit of an art form. I’ve been told that it is dark magic, black box engineering or something to that effect.  We geotechs smile at that, maybe even revel in it. 

We have many variables and initial conditions to determine.  Some of these you really have to drill down to find out (sorry can’t help a drilling reference!), such as your client motivations and needs, why the Architect is designing such a complicated structure, why the Contractor has a fast-track schedule, etc.  We can partner with the AI to understand these human needs.

 

Conclusion

AI and robots may create opportunities for geotechnical engineers and the existing and future workforce, as this CEO envisions. As retailers, manufacturers and even farms become automated, it seems as though we’ll have to keep an eye on what’s next and how we can position ourselves.  We certainly don’t operate in a vacuum, and our competitors are always pushing us.

Without question geotechnical engineering will be a different profession as time goes by.  I expect fewer people will be needed for the field and laboratory data collection areas of the business.  CAD operators will probably go by the way side.  But for the good geotechnical engineers, my hope is that we will still be behind the curtain,  pulling the strings.  

What are your thoughts? Send us your feedback[VM1] .


 

PrintView Printer Friendly Version

EmailEmail Article to Friend

References (8)

References allow you to track sources for this article, as well as articles that were written in response to this article.

Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>
« Leading Thoughts: 6/27/18 Utility Lines and Geopier® Ground Improvement | Main | Leading Thoughts: 3/19/18 Ground Improvement Transcends "Aggregate Piers" »