Children Learn Like Scientists, Fostering Variability and Critical AI Use

Key Quotes

"when i started this project a lot of the philosophers of science said well you know kuhn showed that there was nothing systematic you could say about that it's just sociology but interestingly during the time that i've been working there's been this real change in the way that people think about philosophy of science and we have some some good computational models of how scientific theory change works and it turns out that those apply to children as well"

Alison Gopnik explains that while early philosophy of science suggested no systematic way to understand scientific theory change, the field has since developed computational models that are applicable to how children learn. This indicates a convergence of understanding between scientific and child development processes.

"how do we ever get from the data to the theory and one subcategory of that is how do we ever figure out causal structure which is so important in science how do we ever figure out what causes what just from a bunch of data that we have and what's happened is that philosophers of science and computer scientists have found some systematic ways that you could talk about that and scientists i think mostly you know not necessarily consciously but just as part of what they do and little kids are looking at data and systematically figuring out what kind of structure out there in the world could have caused this pattern of data"

Gopnik highlights the fundamental scientific challenge of inferring theory and causal structure from raw data. She notes that philosophers of science and computer scientists have developed systematic methods for this, which are also observed in the practices of both scientists and young children as they interpret patterns in data.

"but when you look at their actual practice what you see is that in fact kids for example are bayesian and so are scientists now the thing is that in fact in many respects kids are better bayesians than scientists but a lot of it depends on your prior so if you have a very as they say you have a very peaked prior you have a lot of experience you have a lot of reason to believe that this prior assumption is right then it's rational not to change it when you just have a little bit of evidence you should require a lot of evidence to overturn something that you have a lot of confirmation for"

Gopnik argues that both children and scientists operate in a Bayesian manner, with children often exhibiting superior Bayesian reasoning. She clarifies that the effectiveness of Bayesian inference depends on one's "prior" beliefs; strong prior assumptions, based on extensive experience, rationally warrant resistance to change with minimal new evidence.

"the strategy that you see in computer science is this annealing is start out with this wild crazy out of the box high temperature kind of search through the space and then cool off and just fill in the details and you know if you think about your four year old who do they sound like do they sound like the creature that's just moving a little bit or do they sound like they're noisy and bouncy and random and doing all sorts of weird things well the four year old seems to be a really good idea of this kind of random search"

Gopnik draws an analogy between scientific exploration and "simulated annealing" in computer science, describing it as a process that begins with broad, "high-temperature" exploration and then narrows down. She posits that four-year-olds, with their seemingly random and varied actions, exemplify this initial phase of broad, "random search" for understanding.

"and i think the we have good reason to believe that some kind of apprenticeship intuitive apprenticeship model is how you develop those skills so you do something that you think is going to be important you have a teacher who gives you feedback the teacher shows you how the examples of the skill and that kind of interaction sometimes you know these are sometimes the teacher could be quite mean about telling you when you've done something wrong i think that's a really good way for school aged children to learn"

Gopnik proposes that for school-aged children, an "intuitive apprenticeship model" is an effective learning strategy. This involves active practice, receiving feedback from a teacher who demonstrates the skill and corrects errors, which she believes is a robust method for skill acquisition.

"my view about generative ai and i've actually written about this in a paper in science with henry farrell i think you know yes i know henry the political scientist and james evans who's a sociologist i think our whole again our kind of intuitively lay conception of what how ai works is really misguided so we very much have this kind of gollum view about here's this non living thing that we've given a mind to and you know that always works out badly and it's going to either be for good or for ill it will be super intelligence that's kind of the the narrative we think the right narrative is to think of it as what i've called a cultural technology"

Gopnik challenges the common "Gollum view" of AI as a sentient, potentially dangerous entity, arguing instead that generative AI should be understood as a "cultural technology." She suggests this perspective, developed with colleagues, offers a more accurate framework for comprehending AI's function and impact.