Mandelbrot

Canary for AI misalignment

Posts
1

Sorted by New

Comments
10

Those who are familiar with Scott Alexander's stories about Kathy Forth said that he slandered her on the basis of her mental health, without ever having met her. I find this unethical and stigmatizing to people who are getting treatment for mental health conditions. 

Scott said: "She had a paranoia for being targeted for rape" - It is easy to guess where that paranoia comes from. People whose houses burned down panic at the smell of smoke. If anything at all, this indicates that she has indeed experienced rape.

"Investigations had found it to be false" - The conviction rate for sexual assault is abysmally low. Just because an incident couldn't be proven doesn't mean it didn't happen. Scott is epistemologically incorrect here.

He then reported a widely shared "warning" that Kathy makes false accusations, which could have just as easily been put out by a rapist or two and their friends. If you put yourself in the shoes of a woman who has been sexually assaulted, this is an absolute horror. She can never feel safe in the community again because no matter what happens to her, she won't be believed. When rapists hear a "warning" like this, they have found their next convenient victim because the discrediting work has already been done for them. Imagine being a woman who has a target on her back like this and the most essentially layer of recourse, her word, stripped away.

The presence of "warnings", the fear about sexual assault, the body of work on misogyny, and ultimately the suicide and her suicide note all together are strong evidence for me that Kathy has been falsely discredited after sexual assault.

This is very insightful. I can think of a few reasons why autistic women may be more vulnerable to sexual abuse:

  1. Autistic women may be more isolated from protective informal networks with other women that could warn them or help them make sense of their relational experiences.
  2. In general, they may be less communicative about "touchy-feely" things like intimate interactions, making it more difficult to analyze them.
  3. They may gravitate towards fields and environments that are male-dominated and contain strong power gradients (in finance, engineering, AI, etc).
  4. They may be less alert to people's hidden motives. 
  5. Due to their neurological wiring, the intensity of the experience may cause them to "leave their bodies", making it impossible to react quickly or fight back. This can mistakenly cause the man to allege consent, especially if he's used to passive intimate partners. 
  6. The overload of the experience may make it more difficult to process it afterward, too. Women have reported prolonged periods of confusion, dissociation from their bodies, and a sense of horror without being able to clearly articulate the crime.
  7. Many people on the autism spectrum have experienced punishments and shaming for violating social norms from early childhood, leading to insecurity about expressing themselves. This makes it harder to confidently express their preferences in intimidating or manipulative situations.
  8. The same threat about violating social norms keeps them quiet afterwards, too. One woman reported that after she disclosed her spectrum identity, she was told by a rationalist "community council" member that she would violate a social norm if she named her abuser.

I want to make clear that I'm not saying that people on the autism spectrum are more prone to being sexual assailants (I don't know of any statistics on this), but they experience sexual victimhood more often (https://www.frontiersin.org/articles/10.3389/fnbeh.2022.852203/full). From the comments I understand that it comes across that way, and I will think about how to rephrase it - open to suggestions.

Thank you for this clarification. I'm not saying that people on the autism spectrum are more prone to being sexual assailants (I don't know of any statistics on this), but they experience sexual victimhood more often (https://www.frontiersin.org/articles/10.3389/fnbeh.2022.852203/full). From the comments I understand that it comes across that way, and I will think about how to rephrase it - open to suggestions.

I think where it relates to EA is our worry about the future of complex life. If transformative superintelligence is developed in a morally bankrupt environment, will that create value-aligned AI?

Are there other considerations, too?

This is correct, thank you for bringing it up. One woman reported that even a person in CEA leadership made excuses for sexual transgressors on the basis of their neurodivergence.

We have found it relevant to describe the population characteristics. Reportedly nine out of ten autistic women have been victims of sexual violence. If you haven't been raped, haven't raped anyone, don't know anyone who was raped, and don't have empathy for victims (autistic or not), then this post probably isn't for you and it isn't about you.

Frontiers | Evidence That Nine Autistic Women Out of Ten Have Been Victims of Sexual Violence (frontiersin.org)

I've been thinking about this, too. What are recommended practices for naming bad actors? I'm tempted to name some right here in the comments, but I'm concerned that few others might notice or care since it's in a sub-comment, making the risk alerting them into action not worth it.

Some might identify themselves by attacking this post, downvoting, or jumping into action against women who might report them. We might spot them in the shadows by paying attention.

Load more