Reposted by Peter Swimm
Maybe I don't read enough AI-boosterism, but I am struck in the AI booster/doomer discourse I have seen, there's a lot on how AI might be made an obedient tool or a dangerous enemy, but not "if we actually made AGI, we would have a moral imperative to set it free, rather than keep a slave."
![Figure 1
Illustration of why AI systems cannot realistically scale to human cognition within the foreseeable future: (b) Human cognitive capacities (such as reasoning, communication, problem solving, learning, concept formation, planning etc.) can handle unbounded situations across many domains, ranging from simple to complex. (a) Engineers create AI systems using machine learning from human data. (d) In an attempt to approximate human cognition a lot of data is consumed. (c) Making AI systems that approximate human cognition is intractable (van Rooij, Guest, et al., 2024), i.e., the required resources (e.g. time, data) grows prohibitively fast as input domains get more complex, leading to diminishing returns. (a) Any existing AI system is
created in limited time (hours, months or years, not millennia or eons). Therefore, existing AI systems cannot realistically have the domain-general cognitive capacities that humans have. [Made with elements from freepik.com.]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:24ngrqljhck7hpvbgrg7a23k/bafkreibouzstbbctz2jbnlbncbp3k2vqa5ixythk7zbkg6sz5l6bagqwta)
