In one of my recent experiments, I performed a static analysis of a cantilever beam with a cantilever arm.
I first ran the calculations in Frilo DLT software and then gave the same task to the Gemini 2.0 Flash and ChatGPT 4.0 AI models.
The results? Both models were able to calculate the support forces, but they had considerable difficulties when trying to plot the moment progression.
What does this mean?
Artificial intelligence is getting better and better at solving engineering tasks. What does not yet work perfectly today could surpass engineers in terms of precision and efficiency in the near future.
Test it yourself!
Here is a prompt that you can enter directly into an AI such as ChatGPT or Gemini 2.0 to perform the analysis:
Prompt
Perform a static analysis of a beam with the following parameters:
Cross section b=25cm, h=45cm. Total length L=8m. Material concrete C30/37.
Bearing conditions: Left end x(A)=0, fixed spherical bearing.
In the middle of the beam x(B)=5m, movable spherical bearing (vertical displacement blocked).
Right end x(C)=8m, cantilever end of the cantilever.
Uniformly distributed load q=25kN/m over the entire length of the beam and cantilever.
Take into account the dead weight of the beam.
To be calculated:
- Support forces on bearings A and B with an accuracy of 0.01 kN.
- Maximum bending moment in the clamping area M(AB) and above support B in the SGN (partial safety factor γ=1.35).
- Draw the moment curve by calculating and outputting the moment every 0.5 m. My conclusion: AI shows great potential, even in engineering. But how much trust can we place in calculations that are not yet error-free? The human engineer remains irreplaceable – at least for the time being.
The big question:
Would you want to live or work in a building designed entirely by AI?
Autor: Bartłomiej Małek