ChatGPT AI is an impressive language model, after reading an interesting test by engineering.com TotalCAE asked ChatGPT some questions we have received on our help desk in the past to see how it performed.
TotalCAE Helpdesk Question #1: What is the effect of a D3KIL with a sw2 with LS-DYNA?
ChatGPT Response: The “sw2” keyword in LS-DYNA is used to control various features and options during the simulation. The “D3KIL” command is used to deactivate and remove elements from the simulation
ChatGPT is incorrect.
TotalCAE Helpdesk Question #2: What is the cause of glibc_2.27 not found?
ChatGPT Response: The error “glibc_2.27 not found” usually occurs when you try to run an executable file that was compiled on a system with a newer version of the GNU C Library (glibc) than the one installed on your system.
That is correct.
TotalCAE Helpdesk Question #3: How do you configure Ansys Electronics Desktop to use the subnet 10.0.10.0/24 for its communication?
ChatGPT Response: A page long non-sensical made up answer. To long to put here but definitely wrong.
ChatGPT is incorrect.
TotalCAE Helpdesk Question #4 : Write a Fluent journal file that prints the interconnect latency and bandwidth?
/file/read-case-data mycase.cas
/file/read-data mycase.dat
/start-timing
/solve/iterate 1000
/mpi-info
/exitChatGPT is incorrect.
TotalCAE Helpdesk Question #5: How many Abaqus tokens are required for a 32 core job?
ChatGPT Response: Assuming you are using Abaqus/Standard and each core requires one token, then a 32 core job would require 32 tokens. However, if you are using Abaqus/Explicit, which requires two tokens per core, then a 32 core job would require 64 tokens.
ChatGPT is incorrect.
Overall, ChatGPT got one out of five questions from our help desk correct. ChatGPT is a language model, not an Oracle, so this outcome is expected. ChatGPT is not replacing our talented IT CAE ninjas any time soon ( and rest assured, a human wrote this blog post.)
