Ai and gui's

[monkyyy]

llm's/ai will likely not generalize past text. Going to video is 
a big step up in data size and I would argue the success we see, 
whatever your opinion on that is, is embedded knowledge in 
English(if you want to debate chinese is a better language for 
thinking, your wrong but no one should be arguing js is peak 
human language) how many works of shakespear could be compressed 
in the space a single second of 4k tictok dance?

I also believe the ball of hallucinations need to be paired with 
versifiers; with code, protein folding and code we have such 
versifiers and such ai is doing well. But our verification tools 
for guis are rather poor compared to an built in unit tests for a 
range.

The industries answer to this is what I call "dom-walkers" such 
as antigravity; they are kinda shit still and hate the 
implication that being of even further degrading of native apps 
to be "everything is a sassy webpage" and not to lie, I get some 
results that way.

I dont want to go down this path, I dont want shit web apps, and 
d probably isn't ready anyway. Whats a different way forward, how 
can ai write gui apps in d?