Tag Archives: anthropomorphism

text | wondering, The field of AI

<< Where have all the Humans Gone? Long Time Passing>>

—— (dis)Location of (dis)Information ——

Are there three orders of what we could label as “Location Jailbreak?” While seemingly unrelated, these orders of Location Jailbreak could be a category of workarounds that enable humans to confuse other humans about:

  1. (dis)location-information,
  2. location of (non-)(-dis)information,
  3. locating a human,
  4. locating the humane or
  5. locating what is or isn’t human?

Most graspable, present-day instances could be labeled (with a wink) as a Second-order Location Jailbreak such as this one: Google Maps Hacks : lots of “humans” on location, and yet where have all the humans gone? I think to categorize virtual private networks within this order as well. Instances such as discussed in the 3rd reference here below could fit this set too: “fooling social media platforms AI recommendation algorithms by streaming from the street near rich neighborhood… fake GPS…” (Polyakov 2023): lots of humans but where has the location gone?

The First-order Jailbreak (with a wink): the digging of a physical tunnel out of an enclosed location, resulting in dirty fingernails and the temporarily assumed continued presence by unassuming guards which then is followed by erratic gatekeepers once the dis-location has been unveiled. Lots of humans at one location but where has that one gone?

The Third-order disturbance of “location” (again with a wink-of-seriousness) could be at a meta-level of human perception and of loss of ability to accurately locate any human(e), due to the loss of “truth,” destabilized sense of reality, and the loss of the historic human-centeredness (on a pedestal): an example is our collective reaction to “DAN” or “Sydney,” or other telepresenced, “unhinched” alter-ego’s dislocated “in” LLMs & triggered via prompt-finesse / -manipulation / -attacks. This order of confusion of location is also symbolized in a *.gif meme of a Silverback gorilla, who seems to be confusing location of another gorilla with his reflection in a mirror. The LLM (e.g., many an iteration of chatbots) is the human mirror of meta-location confusion. Lots of dislocations and (de)humanization, and where has the humane gone?

Here, in this Third-order Jailbreak, we could locate the simulacra of both location and of the human(e).

—-•
Sources

1.
Introducing the AI Mirror Test, which very smart people keep failing

2.
The clever trick that turns ChatGPT into its evil twin

3.
A February 2023 post on Twitter via LinkedIn and Alex Polyakov

4.
an interesting take via Twitter on The Verge article covering the AI Mirror Test

5.
My first consideration on Twitter of the three orders of jailbreaks.

6.
Naomi Wu’s post on Twitter as referenced by Alex Polyakov on LinkedIn

7.
Simon Weckert reporting on the Google Map Hack.

text | wondering, The field of AI

<< Not Condemning the Humane into a Bin of Impracticality >>


There’s a tendency to reassign shared human endeavors into a corner of impracticality, via labels of theory or thing-without-action-nor-teeth: Philosophy (of science & ethics), art(ists),(fore)play, fiction, IPR, consent & anything in-between measurability of 2 handpicked numbers. Action 1: Imagine a world without these. Action 2: Imagine a world only with these.

Some will state that if it can’t be measured it doesn’t exist. If it doesn’t exist in terms of being confined as a quantitative pool (e.g. data set) it can be ignored. Ignoring can be tooled in a number of ways: devalue, or grab to revalue through one’s own lens on marketability.

(re-)digitization, re-categorization, re-patterning of the debased, to create a set for remodeled reality, equals a process that is of “use” in anthropomorphization, and mechanomorphization: a human being is valued as datasets of “its” output, e.g., a mapping of behavior, results of an (artistic or other multimodal) expression, a KPI, a score.

While technology isn’t neutral, the above is neither singularly a technological issue. It is an ideologically systematized issue with complexity and multiple vectors at play (i.e. see above: that what seems of immediate practicality, or that what is of obvious value, is not dismissed).

While the scientific methods & engineering methods shouldn’t be dismissed nor confused, the humans in their loops aren’t always perceiving themselves as engines outputting discrete measurables. Mechanomorphism takes away the “not always” & replaces it with a polarized use vs waste

Could it be that mechanomorphism, reasonably coupled with anthropomorphism, is far more a concern than its coupled partner, which itself is a serious process that should also allow thought, reflection, debate, struggle, negotiation, nuance, duty-of-care, discernment & compassion?

epilogue:

…one could engage in the following over-simplifying, dichotomizing and outrageous exercise: if we were to imagine that our species succeeded in collectively transforming humanity (as how the species perceives its own ontological being) to be one of “we are best defined and relatable through mechanomorphic metaphors, relations and datafying processes,” then any anthropomorphism within technologies (with a unique attention to those associated with the field of “AI”) might be imagined to be(come) easier to be accomplished, since it would simply have to mimic itself: machine copies machine to become machine. Luckily this is absurd as much as Guernica is cubistically surreal.

Packaging the above, one might then reread Robert S. Lynd’s words penned in 1939: “…the responsibility is to keep
everlastingly challenging the present with the question: But what is it that we human beings want, and what things would have to be done, in what ways and in what sequence, in order to change the present so as to achieve it?”

(thank you to Dr. WSA for triggering this further imagination)

Lynd, R. S. (1939). Knowledge For What?. Princeton: Princeton University Press