uncapybarable5d ago
Parsingphase

GenAI's real alignment problem: We don't need its solutions and we can't handle its intrinsic problems

https://parsingphase.dev/tech/LLMs/theAlignmentProblem.html

A work in 4000 words of progress. Turns out there's no concise way to catalog this mess.

GenAI has an Alignment Problem - we don’t need its solutions and we can’t handle its intrinsic problems

GenAI has an Alignment Problem

parsingphase.dev
Feb 15 at 8:29pmWeb
Show threadAlex WilsonFeb 15
@parsingphase I like the structure of this. This issues are so plentiful and the hype is such a gish gallop of nonsense that no one argument cuts through everything.
Show threadParsingphaseFeb 15

@alex Thank you! It was certainly exhausting to write, and definitely incomplete.

It is so bizarre that it's so difficult to effectively critique the "AI" industry precisely *because* there's so many problems.

Show threadEveryday FreethoughtFeb 16

@parsingphase

I think there is a typo:
"Anyone who thinks they will solve cold fusion or climate change is fundamentally understanding every aspect of the technology."

Shouldn't that be "...is fundamentally misunderstanding every aspect of the technology"?

Other than that, great essay so far.

Show threadParsingphaseFeb 16
@EFreethought Updated, thank you
Show threadKim Crawley 😷 (she/her)Feb 16

@parsingphase

Please check out https://stopgenai.com

Stop Gen AI – Mutual Aid and Political Activism

Show threadJannFeb 16
@parsingphase The economics angle on LLM can’t be emphasized enough.
Show threadJonathan Schofield5d ago
@parsingphase bookmarked. Thank you
Show threadTaran Rampersad5d ago

@parsingphase We seem to be aligned.

And this brings to mind the stochastic penal colony paper by @apparebit in the context of social media.

And looked at from most angles, what is technology based on semaphoric attractors?

Another stochastic penal colony. Busy right now, but his paper I referred to is in this link:

https://knowprose.com/2026/02/the-building-of-the-stochastic-social-media-penal-colony/

Now I guess I need to do one on AI.

So much writing to be done, and the targets keep moving...

Show threadTaran Rampersad5d ago

@parsingphase @apparebit Interestingly, if we apply the stochastic penal colony to what we're all seeing (the elephant octopus?)...

And toss in this...

https://knowprose.com/2026/03/why-ai-writing-converges-the-adjacent-possible-of-language-models/

The only real aspect of stochastic would be... for the users...

I need to percolate that.

Keep writing!!!!

Show threadAleksandr Koltsoff5d ago
@parsingphase A wonderful work, thank you!
Show threadMark Lynch4d ago
@parsingphase thank you for sharing that. It’s an absolutely brilliant read which very eloquently expresses pretty much everything I feel, but sometimes struggle to articulate without ranting, about GenAI
Show threadmatt.clarke4d ago
@mark_f_lynch @parsingphase
+1 to this - thank you. I also struggle to explain my objections clearly, partly because there are so many they all want to come out at once.