01072022 by Lenny:

眼下五大洲各地本已根深蒂固的多维度不平等在疫情的催化下愈发不可逆地恶化;越来越多人在各自的生存处境中摇摇欲坠,成为需要〞激进关怀”的对象。这是不假的事实。我不禁想问勤更豆瓣勤录播客的某些博主们(票圈里有的话,这是给你们的问题):在当前世界经济制度无法被大规模改进的情况下,作为普通人,我们到底如何以非激进的、遵循社会约定成俗和公共理性的方式关怀身边正在坠落的任何人(不分种族、国籍、宗教、性别身份、教育背景)?如果没有确切答案,我们是否应该重新审视”阶级政治的时代已经翻篇、不应重提”这种自我蒙蔽式的论断?If an”elevated discourse”(consisting in preferred ideologies andetiquettes) is all that you care about, howcan you still associate yourself withconcepts as multi-layered, profound andentrenched as”decency” and “empathy”?


02222022 on Twitter:

Venom is an example of a villain who evolved into a hero because he was popular with readers and writers decided to rehabilitate him. By contrast, Magneto evolved into a hero because every decade since Reagan it’s harder and harder to pretend he’s wrong


@bayesianboy on Twitter, Jul 11, 2023:

It seems to me that at least half the people who believe in the in-principle prospect of something in the ballpark of our current AI/ML technologies exhibiting agent-like behaviour do so because they have fundamentally misunderstood the point of intentional stance-taking.

If you’ll all allow me a moment of insufferableness I will assume briefly the role of Dennett-exegete. The bulk of Dennett’s life’s work, what he will be remembered for, is his treatment of mental kinds—phenomenal consciousness, free will, teleology, selfhood, theory of mind, intentionality, cognition itself. Mental kinds such as these have long been associated with an ineffable essence only accessible from within; as such, our philosophical and scientific purchase thereon has been stymied. Dennett’s insight, time and again, has been to put forth conceptual analyses of such terms identifying them instead with their outwardly perceptible manifestations, with what is discernible to a third party observer. This serves the crucial function of rendering mental kinds amenable to empirical investigation. As such, Dennett members among those who have been both celebrated for and accused of handing philosophy of mind off to the neuro- and cognitive scientists.

Many have interpreted Dennett’s work, especially his notion of the intentional stance, as implying that, for mental kinds, a simulation of the phenomenon is tantamount to the phenomenon itself. This mistakes the intended upshot of the work. It is not sufficient for identification or assimilation that someone might, in a highly artificial setting, over limited interaction, or with heavily constrained access to the system’s operation, be unable to differentiate it from the phenomenon. To be assimilated under a mental kind concept, a system must reproduce the functional form of the behaviour in question, not mere superficial tokens thereof.

I see also a general proclivity to mistake theses intended as research heuristic, as a definitional ansatz which renders empirical study possible, as statements of deep metaphysical truths, when the empiricist spirit in which they were intended abjures idle metaphysics.

Nothing like AI/ML systems in present form are capable of functional or behavioural replication of mental kinds in the sense necessary to attribute them with mentalistic faculties, even heuristically. The primary concern in the development and adoption of these systems is and will remain—as Dennett has long stressed—how we humans interact with these systems and wield them as tools. The more AI tools are designed to superficially approximate the hallmarks of mentalistic properties, the more limited cognitive agents such as ourselves will over-attribute competence or comprehension to these systems in ways that will have dire consequences for their use.

Ultimately, we do not want “artificial intelligence.” We do not want systems designed to replicate human judgement, behaviour, or labour. We want tools that are amenable to effective human use and which, when used by humans effectively, offer an improvement over the status quo.