July newsletter! Featuring @retrievelearn.bsky.social and Bethany Lisi, some new curiosity & learning research, why college should be more fun, a battle about phones and play and mental health, and one solution for what to do about all that zucchini. newsletter.sarahrosecav.com/on-curiosity/
On Curiosity
The end of July is always takes me by surprise. Even more so since making the transition from faculty to educational developer, where August is one of the busiest months of our calendar year (orientin...
newsletter.sarahrosecav.com
![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/bafkreie5dbxbxpaszqwfbgne6pi7kf26im6chq45olvzxyyrg7eitq6uue)
