|
I recently had a few discussions with a person who strongly believes that large language models are already generally intelligent, and that a so-called ASI (Artificial SuperIntelligence, IIUC) is overwhelmingly likely to emerge within the next few years. These beliefs are founded in (imo) breathless, grandiose, generally unfounded readings of essentially every AI paper posted on the arXiv, but when others point out that said conclusions are not [yet] supported in practice (i.e. LLMs cannot yet replace the average software engineer), this person retorts that LLMs' performance on synthetic benchmarks is the only thing that matters, and that (e.g.) a practicing software engineer with decades of experience is utterly incapable of evaluating the merits of a particular LLM as a programming tool. Long story short, I maneuvered this person into a rhetorical corner where they had no choice but to abdicate or say to my (virtual) face that a master carpenter is literally incapable of evaluating the practical merits, or lack thereof, of a particular hammer that they have held in their hands and used. That is when I decided to stop having substantive conversations with this person, because there is no arguing with such an absurd depth of self-serving epistemic nihilism. I was satisfied with simply proving that it was there, to myself and others in the conversation. --- > "But the reality is, it is safe... just with conditionals, which I mentioned, and with instructions to child, and of course taking into account the specific child too" You are offering broad, utterly worthless generalities with a side of BS hand-wringing about politics (which this is not). I am offering concrete scenarios and questions, because each child is real and they live in real physical environments and any injury or death they sustain in the course of a dangerous activity is real. So I ask you again: is it appropriate for me to let my nine month old crawl around in the street? Why or why not? Is it appropriate for me to let my four year old use the sharpest chef's knife in the drawer? Why or why not? I would like you to commit to answers here, to establish a baseline, so that I know you are not so steeped in self-serving epistemic nihilism that you would tell me to my face that a carpenter cannot say if a hammer is a good hammer or a bad hammer just to salve your own wounded ego. Once you've provided those answers, let's try a less obvious (I hope!) scenario. First, at (approximately) what age would you say that a child living in a sleepy subdivision with wide streets, long sightlines, and minimal traffic should be allowed to tool around on a bike in the street and visit her neighborhood friends living within a handful of blocks? Now, having answered that, how would you answer it for an identical child living in a high-rise condo in downtown LA? Are they the same age? Is your answer simply "as soon as they are able to ride a bike", taking judgment entirely out of the equation? Or will you admit that all things are not equal? And finally, assuming you do admit that parental judgment has a role in keeping children safe, what exactly is it that makes you believe you are more fit to judge the cycling safety landscape of my neighborhood, in my city where I have been cycling for almost two decades, than me? Without even knowing what neighborhood in what city we're talking about, no less! I look forward to reading your answers to these questions. If you cannot answer them I will have to assume you are participating in bad faith, and this will be my last reply in this thread. |