I and yes, I get it, use claude, pay for gpt pro, use glb glob, whatever, I never go to buy a sweater that doesn't fit and the lady tells me "oh but if you take it home after you pay for it it will fit right away"
What I find absurdly annoying is that for complex code, there is no rhyme or reason: it writes no main, the main line of code is spread between functions, assumes initial conditions, makes up gpib commands that don't exist and not a single "try" in sight
I want to understand if it is common practice in python to define functions as you go linearly down the code.
Like:
* Declare variables
* initialize something
* next set of functions
* some other common core cross coding
* MORE FUNCTIONS
* main
* ADDITIONAL FUNCTIONS
First: Chatgpt generates ok code if you have to do one thing only. Like a sum or a division. Anything remotely complex OR god forbid requiring scipy good luck during iterations.
Anything outside python I have tried: LabTalk, LabView, direct programming to an OSA, forget about it.
I've been "vibe coding" as the kids say in the past month because I cannot be bothered by some small programming tasks that I had read were good for AI agents.
Here are my takes on vibe coding from a not so good coder, that end up rediscovering known code good practice.
in a time when being an immigrant feels like a moral failing, I felt so much during that half time show
Re: linkedinspeak
a few words that should not be used if you are not in linkedincore
thrilled
fucking stupid vacations
I started a downward spiral because I cannot attend a concert I want to go to, because I have to (stress on have to) take vacations later in the year and they collide with a concert for a band I don't even like so much in the first place
one of the worst things that can happen to me are vacations (holidays, for the british anglicized nerds out there) because I start thinking, and the worst thing that can happen to me is thinking because I can't sleep (see: time)
I should make a thread about "the plot that took me almost a year to get and it does not even look that great"
que salga esa plata
the tyranny of stem o que ase
Damn it LinkedIn why do you send me messages if I know someone as if they invited to link with me yet it was a suggestion?
Now I will look like a terble person for switching out grr.
Nonetheless I catch a lot of primarily younger tiktok influencers delving into topics that they are clearly not experts in and giving sometimes off, very often incorrect statements about fields they are not experts in which bothers me. Particularly optics explanations from non-optics experts.
Angela Collier, a great science communicator with a small-ish YT channel talked about this in a video already: The Gell-Mann Amnesia: youtu.be/wBBnfu8N_J0?... and puts a lot of my thoughts in a great way mostly on my discomfort with off-comments, and I have to remind myself that I am not the target
I've had that in mind for a long while.
There is a section of scicomm mostly on Tiktok that bothers me slightly because it is clearly scicomm influencers with adjacent scientific knowledge to a topic, but the explanations are always slightly wrong.
very interesting! I was not aware that confocal microscope coupling was possible. I tried with an inverted microscope for a few weeks but I could not figure out the back coupling from the emission port.
facebook marketplace was, for a minute there, a legitimately good tool to just buy and sell stuff in your area with some degree of safety since the sale is linked to social media profiles so you could more-or-less tell true good deals
not even 2 years in is weird scams and back to being useless
Also: HR does not exist in academia. Discussions, mediations and agreements are done in closed doors, and being in good terms with everyone, included non-technical staff is crucial to avoid bad blood and active sabotage. Be cool, be helpful, but set limits and you may be fine.
very nice sum-up! I am doing my PhD later (am 36) and something I constantly see and only came to terms after industry experience is that office/lab politics are important, and being respectful and mindful of the internal dynamics and hierarchies is important.
photolithographic behavior of resist under exposure generated by AI
I mean me quieren ver la cara de tonta or did you really think no one would notice the sudden change of illumination parameter
I had the misfortune of receiving a lab report with some clear evidence of AI-LLM output in the core text. For the most part I am not too bothered because a) brother same, i hate pointless work but b) I specifically told the students to keep the report short so they could actually report issues.
my git repo tracks 2 things: Novel 2 which i am copy/editing and my thesis for the phd which i am adding meaninglessness to (as my core experimental work starts once my snitride wafers arrive) i hope some managers looks at it and says "oh yeah got boy here"
there was some "thermal action" on my maracas and they spontaneously got detached from the wooden handle, waifu claims "me ves cara de que yo estoy agarrando tus chingadas maracas" but spooky action-at-a-distance seems less feasible than petty domestic revenge