disclaimer: i'm trying to keep a neutral or happy tone in most of the writings but there's a bit of negativity in this one in particular.

i've been hyped about ai since forever but starting from 2023 i've been a bit disillusioned with the state of llms, gen ai, etc... here's a list of things i find annoying about them:

• nobody has come up with a solution to llms' inability to distinguish real information from hallucinations
• chatbots have been lobotomized (this is the fault of corporate decisions, not ai). i still remember the playground before openai started adding all their guidelines. the way everything now is safe, toned down, friendly, unproblematic... is ultimately anti-art and anti-human.
• everything is closed source and proprietary. there have been attempts at open source projects, and the big companies sometimes release public weights (after lobotomizing them) but most of the real thing happens behind closed doors.
• 99% of gen ai's contribution to humanity so far has been making it easier to scam people, fake evidence, cheat in schools, write wedding speeches... and litter the internet with petabytes of garbage images and misinformation.
• they are frankly not impressive at solving new problems or coming up with new ideas!
• ai generated images just aren't great (yet?). they all have that same feeling of "slop" that anyone who has looked at enough output of dall-e or SD can recognize in a second. or maybe it's post hoc idk.
• they are EVERYWHERE. every software site has something about ai in their homepage. every program has an ai button. almost none of them are real use cases. we don't need any of this.

2026 update: looking back on lalamo