But that would mean doing less, and that's by default bad. We must take action! Think of the children!
I tried at my workplace to get them to stop mandatory rotation when that research came out. My request was shot down without any attempt at justification. I don't know if it's fear of liability or if the cyber insurers are requiring it, but by gum we're going to rotate passwords until the sun burns out.
I wonder if that difference in mentality is a large part of the pro- vs anti-AI debate.
To me the AI is a very smart tool, not a very dumb co-worker. When I use the tool, my goal is for _me_ to learn from _its_ mistakes, so I can get better at using the tool. Code I produce using an AI tool is my code. I don't produce it by directly writing it, but my techniques guide the tool through the generation process and I am responsible for the fitness and quality of the resulting code.
I accept that the tool doesn't learn like a human, just like I accept that my IDE or a screwdriver doesn't learn like a human. But I myself can improve the performance of the AI coding by developing my own skills through usage and then applying those skills.
To summarize the Wikipedia article on linguistic relativity, the "strong" hypothesis that language determines thought has been debunked. But there are many things that a language influences. To use a computer analogy, all mainstream programming languages are Turing complete, so you can express any computation in them. In this sense the language does not determine what programs you can write. But in practice, as any computer person will tell you, different languages are good at different things. And that is kind of this paper, they cite a lot of examples where English has poor vocabulary or odd quirks, and show by comparison to other languages that this measurably affects conclusions about certain cognitive abilities. The issue they're complaining about is like if you benchmarked Python programs and tried to draw conclusions about the speed limits of computing, but never tried C++ or assembly.
Tell it to the papers: https://www.cs.drexel.edu/~csg63/publications/onward24/onwar...https://arxiv.org/pdf/1808.03916 As vague analogies go, much more ridiculous and vague things have been published and peer reviewed and even gotten significant citations. Like ecological niches and invasive species, DNA as genetic blueprints, selfish genes, ... About all that can be said about these is that they are closer to the truth than what came before, and that if you actually learn the field then you can appreciate how they kind of get it right.
Yet what about aphasia? There seems to be direct evidence that thought, action, language are in conflict rather than seamless, so language plays a weird role in that some of us are programmed and others are blissfuly unconnected to their effects.
If aphasia is evidence that some of us don't use language to think, then language is nothing more than a programming language.
Whatever programming language using language is irrelevant to people with aphasia.
Typically people develop aphasia after a stroke. This provides a natural before/after comparison. When for example Mark Brodie had a stroke, he wasn't able to speak - and he also wasn't able to read or write computer programs. This indicates that natural and programming languages use overlapping regions of the brain.
Regarding language being in conflict with other modes of thinking, there is indeed a "savant" effect of aphasia, where after an aphasia-inducing stroke people suddenly develop amazing abilities in visual art, music, or mathematical thinking. But it is not consistent - most people don't develop these. And it comes with impairment of emotions, memory, etc. So really what the evidence suggests is that some people suppress parts of their brain, and these injuries unlock that potential because the suppression mechanisms break. It's most likely a cultural thing - people act how they are "expected" to act. There is some evidence that females actually have more biological capability for math (in cultures with very high gender equality) but typically you see less performance, so the conclusion is that the culture essentially "programs" in the lower performance. It is probably the same with the supposed "conflict" between thought, action, and language - the culture treats these as distinct modes and inhibits cross-modal thinking like synesthesia.
Sure, but thought does not end or get impeded in aphasia, merely the ability to use language. People can reason, gesture, react, which tells us they're either divergent or entirely separate abilities. Any language is an external program that runs a cultural control system, not a communication system that directly connects mental states.
You can't generalize like that. Aphasia is a symptom - of course by definition it is merely the inability to use language. But people with stroke will typically show a lot of impairment, not only aphasia - they will have difficulties reasoning, gesturing, reacting, etc. Different people will have different abilities and different levels of impairment, but this tells you very little - it is not like each ability has its own little neuron in the brain, fMRI has confirmed that most activities involve several different parts of the brain. There are complex thoughts that require linguistic involvement to process, sign language and dancing combine gesturing with language, etc. The main thing the OP paper shows is that language is pervasive and intricately involved in how activities map to mental states.
Yeah, so what that paper shows is that 6-7 cognitive tasks show low involvement with the language areas. What it doesn't show is that all cognitive tasks are independent of the language areas. As the paper itself admits, some forms of reasoning seem to involve language.
It demonstrates what divides the brain, actually, if you read her other papers and public layperson statements as well where she claims we don't use "language to think" as mental events are specific and language is arbitrary.
This seems like the (broadly accepted, AFAIK) weak form of Sapir-Whorf (language has impacts on cognition) but not the (generally viewed as debunked) strong form (language places strict limits on the bounds of possible thought).
I'm actually on the fence about strong Sapir-Whorf now. It's rather suss that machines started exhibiting some form of <reasoning> capabilities the minute they could successfully parse natural language, and not a moment before.
My company is paying for Claude Max for me and a dozen other developers. The others are using the API. If their API usage cost hits a level where it's cheaper to move them to Max, they're moved to Max.
There's no hard mandate to use Claude Code, but the value for us is clear to exec management and they are willing to foot the bill.
Why would it be any more "hell" than the kind of science practiced by natural philosophers? You can still do science on things you don't fully understand.
reply