Timnit Gebru (she/her).May 30, 2023
Its like a coordinated DDOS PR attack from these existential risk people and they're succeeding in making me constantly talk about them.
Arcane SciencesJun 1, 2023
Show threadSasha Costanza-Chock

@timnitGebru maybe we should start a catchy label abbreviation to change the frame?

They are x-risk, we are

A-risk (actual risk?)
R-squared (real risk?)
R-harm (real harms?)
X-harms (eXisting harms?)

May 30, 2023 at 8:12pmWeb
Show threadTimnit Gebru (she/her).May 30, 2023
@schock I love this.
Show threadSasha Costanza-ChockMay 30, 2023
@timnitGebru which one should we try? Or another snappy term that's better
Show threadSasha Costanza-ChockMay 30, 2023
@timnitGebru x-harm? As in eXisting harm?
Show threadSasha Costanza-ChockMay 30, 2023
@timnitGebru Reducing x-harm (actually eXisting harms) is a more powerful, ethical, and in the long run, effective approach than speculation about so-called x-risk.
Show threadTimnit Gebru (she/her).Jun 1, 2023
@schock I’m a fan of all of them really.
Show threadGilberto FicaraJun 1, 2023

@schock @timnitGebru problem with long-termist crazyhats is that they don't actually believe in reducing x-harm

How do we fight that?
Or we don't engage them at all?

Show threadSasha Costanza-ChockJun 1, 2023
@zompetto @timnitGebru i think we just focus on developing a stronger alternate narrative
Show threadmelanie ensign (she/her)Jun 2, 2023

@schock @zompetto @timnitGebru I love the thought behind the creative terminology. I do worry it’s too complicated for media & other potential messengers. The suggestion to developer a stronger counter narrative is smart.

The reality is existential threats are being used as a distraction so people don’t pay attention to the harms happening right now. The individuals behind this movement are hoping to neutralize oversight by the public & regulators with fear of the boogeyman. Fight, flight, or freeze — just don’t look at the man behind the curtain.

Show threadNomiJun 2, 2023
@schock @timnitGebru
X-harms are more urgent than F-risks (future), since those are risks which may or may not even come to pass.
Show threadCorbinMay 30, 2023
@schock @timnitGebru "harm" is a great word to bring to the table. Maybe "E-harm" would work, talking about empirical/extant/evidenced harms which -- unlike all-or-nothing black-swan apocalyptic contemplations -- actually do obey the basic rules of probability. This would be something that we could then take back to the workplace and use in discussions of service reliability and quality.
Show threadGuy EmersonMay 31, 2023
@schock @timnitGebru How about "y-risk". Like, why-the-fuck-are-you-building-this risk.
Show threadAndrew HundtJun 1, 2023
@schock @timnitGebru a-harms actual/addressing/anti-harms
Show threadAndrew HundtJun 1, 2023

@schock @timnitGebru or flip the script justice-[TBD].

Which I guess is the AJL’s approach?