[FRIAM] GPT-3 and the chinese room

[uǝlƃ ↙↙↙]

IDK. The controversy over the book "White Fragility" (e.g. https://www.theatlantic.com/ideas/archive/2020/07/dehumanizing-condescension-white-fragility/614146/) is a modern refinement. But our extended conversations about steelmanning, active listening, and demonstrations of understanding dovetail (as I tried to say to Gillian in response to his rant about visitors to Santa Fe).

No matter what we do in the isolation of our armchairs, it's irrelevant. Nick's invocation of embodiment is great. We *are* deep memory functions of our environment. But there's a next-order out, a 2nd order composition where interaction with *other* deep memory functions is qualitatively different from interaction with our environment. (I'm including dogs as deep memory machines, obviously ... so interacting with dogs counts. 8^)

On 7/28/20 11:18 AM, Jon Zingale wrote:
> From my perspective, it is helpful to consider a larger history of the
> consciousness debate. In what could be considered /the beginning of the
> end/ of African slavery in the west, natural philosophers would seek to
> find in the physiology of black men structures to explain their /inferiority/
> and in some extreme cases their /inhumanity/. Neal Stephenson, in his /Baroque
> Cycle/ (~)trilogy, caricatures /how embarrassing white people //be/ when in the
> pursuit of the consciousness question. In the novel, a Nigerian born
> linguist named Dappa attempts, through the power of words and ideas, to
> argue for the freedom of enslaved people in the west (circa the 1660s).
> Meanwhile, many spectators look on and muse about how nearly human Dappa's
> arguments sound, but ultimately /must not be confused/ with the utterances
> of a /reasoning being/. Stephenson highlights the cruelty and tyranny that
> enterprises like these can create. The Turing test generally reeks of
> this sort of pursuit. Rather than beginning with the assumption that the
> other experiences and then wondering what that experience is like, we set up
> an endless procession of tests which on the one hand we hope will /converge/
> in the limit to understanding, and which on the other will /entertain/ us
> in the meantime. The Turing test has appeared to me to be a horrible
> diversion from the discovery of more promising methodologies. One day,
> perhaps there will be a construction that passes all acceptable Turing
> tests, and this day will be a sad day because we will likely still have no
> answers to our initial investigations regarding what it is to be conscious
> and what it is to /experience as another does/. Perhaps, the question will
> be considered solved for /all time/ and the potential needed to reopen
> the topic exceedingly expensive. Somewhere, Minsky expressed remorse for
> how some of his results in the field of AI had managed to close the general
> AI question for decades, ultimately shifting the pursuit away from the hard
> question and instead towards the pursuit of novel gadgets. In the spirit
> of EricC's comments about the distinction of surface tension and PH, if
> consciousness is a thing, then it should be so whether or not we all agree.

-- 
↙↙↙ uǝlƃ