Maybe 18-24 months ago I tried to use OPEN AI to write a simple program that would allow users to have a nice interface for plotting and looking for features in extremely large .csv data sets (essentially data files 100′s of Mega Bytes in size that contained labeled rows and columns of data that you would normally use EXCEL to plot) but I wanted something more user friendly where you could use your mouse to zoom in and look at features of interest and change quickly what data was being plotted….

It started off promising but became a hopeless task, the AI started giving me incorrect code that needed to be debugged over and over, and once that problem was debugged it had forgotten the other features I wanted the program to have- so when I tried to have it add those features it then added buggy code and ALSO reverted the previous code we had just debugged to the original buggy code- it was frustrating.

I recently made use of OPEN AI to help me write a program to model in a physically realistic way the far field pattern of a laser beam transmitted through an aperture with arbitrary shape, with obscurations and user adjustable wavefront error, and all sorts of physically real feature that calculated the diffraction effects and gave me the intensity distribution of the beam at whatever range I requested… it ended up being ~700 lines, I did it in stages and checked the results against known examples and it was flawless- there were a few errors it made, and corrected them once I pointed them out but the process was easy- fun really… I don’t think I was even coming close to giving it tasks it would find difficult

The advancement rate of AI is staggering- could I have written the program? Probably- given LOTS of time I simply didn’t have, but I knocked out the majority of it with AI in a couple of afternoons (actually the basic framework was one afternoon, after that its just been me adding nice to have features) . And NOW they are starting to allow the AIs to take over the improvement of their own code- a scenario that many experts in the field warn will bring about doom for the human race. You can describe an app you want to create, or a website you want to build and it will do it… I have even started considering using AI to create a backup board for the HZ (If I ever have the free time) even though I have only created ONE crude webpage many years ago and know next to nothing about the code to do such things…

The revolution all this will bring about is staggering- the job losses, the advancements, the ever accelerating rate of AI improvement once slow humans are out of the loop… it is very scary… I love having access to these tools, but longer term- like the next few years- or even months- the implications are terrifying… I don’t THINK my job is personally threatened – but the vast majority of the workforce is in danger- what happens when vast swaths of the jobs can be done faster better and cheaper with AI instead of humans? How fast will we reach a point that AI starts to come up with its OWN goals and decides humans are simply a nuisance?