Raw LLM Responses

Just some examples to understand what I mean with this: 1) "ChatGPT, how heavy is each component in a bus? How many of them (seats etc.) are there? What materials are used? What are the physical properties of all of these (density, heat capacity etc.)? How would they burn (heat release rates)? Etc." This would be a research topic that might require 10-20 materials to be tested in small-scale fire testing and days of research by a team of students, but with AI can be answered (sufficiently good) in a few minutes. 2) "ChatGPT, can you write me a neural network that is able to optimize these parameters A, B, C and D to minimize this number here, generate a script for a fire simulation, run the script and check if the results match to reality? If not: Repeat everything starting from optimizing ABCD again?" Normally, this can require weeks and one guy who logs in remotely to the workstation several times a day, doing calculations, write new scripts with estimated numbers, start these scripts and so on. I would estimate 40 working hours for this. The neural network by ChatGPT is able to do this in something like 2 days AND everything is completely automated. So it needs one guy to start the script over the weekend and check the result on Monday. This is less than 1 hour. 3) "ChatGPT, we like to have a Telegram bot that sends us a message whenever a simulation has finished or crashed" Works perfectly. Normally, someone needs to check regularly (at least daily) to see if something failed or finished, because all this unused simulation time can be used for the next simulations or to speed up projects considerably. .... Many more examples.