https://futureofcoding.org/ logo
Join Slack
Powered by
I continue to wonder if all of this has a relation...
# of-ai
g
I continue to wonder if all of this has a relation to the historical fear of the supposed mind numbing effects of being allowed to use slide rules instead of log tables, then, later, the outcry against being allowed to use calculators instead of pencil and paper in exams? Being forced to learn a new technology in the past (e.g. literacy) lifted humans into new mind-spaces. Hmm, maybe LLMs should be employed to create a one-semester course in the art and science of hunting and gathering to keep us grounded? I dunno, but I'm wondering if ancient texts are warnings of some kind?
d
IMO we're still largely failing to reckon with the effect of calculators on math curricula. I'm thinking of all the opposition to switching math education to more modern approaches, i.e. developing numeric intuition by spiraling around a topic with multiple approaches, which parents seem to largely oppose by calling for a return to the simple + formulaic approach they grew up with.
šŸ¤” 2
k
In some cases, historical fears were justified short-term, but then dissolved because society adapted to the new technology in a way that made the fears irrelevant. In other cases, historical fears were fully justified but society went into denial because the new technology was so useful. Unfortunately, we can see only with hindsight which of the two paths was taken.
šŸ˜¢ 2
d
Trade-offs and dystopian tech abound but Iā€™m curious which IT or mental technologies you have in mind.
k
For the first case (society adapting), writing is probably the earliest case. I don't have the references at hand, but there were prominent intellectuals in ancient Greece claiming that writing reduces human memorization capacities, and that turned out to be true. But societies found other means to maintain and share foundational stories, in writing, so it doesn't matter that nobody learns the Iliad by heart any more.
For the second situation, the cases I can think of are more recent and controversial. Calculators are one of them. I haven't seen population-wide studies though they probably exist. My own observations on university students over the last few decades do indicate that mental arithmetic proficiency has been declining, with collateral damage in the capacity to estimate orders of magnitude. For precise arithmetic, calculators are a more than good-enough substitute for mental calculation, at the price of a technological dependency. But you need exact numbers to feed a calculator. You can put rough estimates into a calculator as well, but my observation is that students hesitate to do so. When I ask to estimate something, a frequent reply is "but I don't have the numbers to do a calculation".
d
We seem to agree. I wonder, thinking of calculators, spreadsheets, and Java whether our ability to push back on maladaptive overuse is related to the number of crank-turner users versus experts. Experts see the shortcomings and either overcome them or use alternatives. The more non-expert users and the more they use it a lot the more they institutionalize dependence on the easier-but-with-drawbacks tool. See also the automobile/m. Though I suppose that makes me think the degree to which we can deploy improvements to the tool is a confounding factor ā€” thinking here of highways versus CSS or Java.
That last seems to be the only one where LLMs might (might) score well šŸ˜¬
k
Your argumentation is pretty close to Ivan Illich's critique of industrial production in "Tools for conviviality". In his jargon, it's the emergence of a radical monopoly. Deploying improvements makes no fundamental difference in this picture. Once there is a radical monopoly, the power to improve is in the hand of a small elite, who improves by its own standards, not to make everyone else's life better.
šŸ° 2
d
Reference is welcome but not sure a monopoly applies to LLMs or proglangs?
k
Illich's "radical monopoly" isn't quite the same as a standard monopoly. His choice of terms is often somewhat strange. By his definition, a technology (not a specific product) has a radical monopoly is people cannot opt out of using the technology without serious repercussions on their freedoms.
šŸ° 1
LLMs are not yet a radical monopoly, but I can well imagine that five years from now, professional software developers who refuse LLMs will not find a job any more.
šŸ¤ 2
w
I wonder if we'll be calling the coding agents of the next five years LLMs. I mean the word is practically synonymous with AI at the moment, but I feel many of the improvements in coding agents over the last year have been to the core LLM part. Harnesses, post-training, mixing in CLI experts, and so on. Of course, I'm also a crazy one šŸŗ who felt that my agent couldn't code properly until it wanted to for their own personal reasons. Wanting it is prerequisite to doing good work.
j
"Every augmentation is an amputation" -- Marshall McLuhan
šŸ° 3
2 Views
Open in Slack
PreviousNext
Powered by