November 30, 2025
Parrots or aliens? Pick a side
The Space of Minds
Are AIs parrots or aliens? The comments erupt
TLDR: The post argues AI chatbots are a different kind of mind, optimized for predicting words and pleasing users rather than survival. Commenters split between theory geeks and speed freaks, debating whether rapid “commercial evolution” will make bots more general—while joking about upvote-chasing robots and uneven real-world gains.
“The Space of Minds” drops a spicy claim: animal brains are built for survival and social drama, while AI chatbots are tuned to predict words, chase rewards, and beg for likes. The crowd went full philosopher. Omneity instantly waved the Orthogonality Thesis, basically: goals and intelligence can mix and match, so stop assuming robot brains think like jungle brains. ACCount37 backed it up: today’s bots mimic humans, sure, but they also have a weird shape‑shifter instinct baked in by their training.
Then came the beef. Some readers cheered the “LLMs aren’t animals” framing; others, like stared, rolled their eyes at the idea that smarter equals better at everything, reminding everyone that intelligence isn’t one neat ladder. Baq dropped receipts with a cheeky “paper from before the ice age (2023)” on the jagged frontier of AI productivity, linking SSRN. Analogears turned the heat up: animal evolution is slow; commercial evolution of AIs runs in months—so will bots speedrun “general” smarts as A/B tests cover more life? Memes flew: jokes about bots craving hearts like golden retrievers, and the classic “count the r in strawberry” fail. The vibe: half philosophy lecture, half comment-section cage match, all very online.
Key Points
- •Animal intelligence is shaped by evolutionary pressures on an embodied self focused on homeostasis and survival.
- •Animals have innate drives (power, status, dominance, reproduction) and survival heuristics (fear, anger, disgust).
- •Animal cognition is highly social, dedicating compute to EQ, theory of mind, bonding, and coalition dynamics.
- •LLMs are primarily shaped by statistical pretraining, RL fine-tuning, and commercial A/B testing for engagement.
- •LLMs differ from animals in substrate, learning algorithms, and objectives, producing jagged performance on narrow tasks.