Evaluating the AI Safety cause area: a two-year plan

My pretending to have readers makes this blog a self-commitment scheme: if I write that I will do something, the imagined possibility of social shaming will make me more committed to actually doing the thing. That is why I’ll line out my plans for evaluating whether I believe the AI Safety cause area as perceived by EA’s is any good.

I used to think the arguments were pretty convincing, and I liked how it made my skills in math and CS super relevant for a morally important thing. But then I listened to Bostrom’s Superintelligence: Paths, Dangers, Strategies and found the book’s arguments thoroughly lacking. There is an asymmetry of passion in the AI risk circles where the believers are hyping the cause area while the non-believers don’t interact with the ideas at all.

The following step plan is based on the assumption that the LW-aligned view of most EA’s is misguided, but that there is important work to be done to make sure that what we are currently calling AI is used properly. Every step in the plan is meant to defend part of my beliefs, and I trust myself to notice when my argumentation is shaky. If my view changes, I should feel free to change the plan to look into my new views instead.

After completing a step, I will write a blog post describing the outcome and related thoughts. I will try to complete at least one step every two months so that the plan is done before you re-enter the job market after the end of your PhD. I don’t have to do the steps in the listed order.

  1. Make a plan listing the steps to think through the AIS cause. Completed.
  2. Uneducated guessing: directly oppose the astronomical waste argument. See if you can collect the necessary entropy to argue that, even if astronomical waste would be astronomically bad, supporting the AI Safety cause is still bad in expectation. Read at least what Bostrom, Beckstead and FRI have written on the topic, maybe more.
  3. Reread Superintelligence. Try to list the major points in the arguments and list objections. Does it actually argue for a Yudkowsian view, or is everyone misinterpreting it and does it secretly argue for a more mainstream view?
  4. Read various online sources to get more of an idea of what the mainstream view is among AIS EA’s. See if you can list research agendas on AIS different from MIRI’s.
  5. Argue that MIRI is not an effective charity regardless of the status of AIS as a field. I am not sure I truly believe this, but they seem so deeply incongruent with the standard academic practice that I should spend some time thinking about them. I kind of expect that this post will feel like punching down.
  6. I think Paul Christiano is a smart and serious person. He co-authored one of the best papers in your field of the past decade so he is not a crank. Read some of his writing on AIS to see if it holds up to scrutiny.
  7. Argue that average utilitarianism is superior to total utilitarianism, and astronomical waste cannot exist.
  8. Argue in more familiar terms why a paperclip maximizer wouldn’t act as some people fear they would.
  9. Argue that the capitalist’s alignment problem, insofar as it is meaningful and solvable, will be solved by the market.
  10. The social democrat’s alignment problem is a meaningful concept.
  11. Educated guessing: redo the entropy gathering from earlier, but now while having more knowledge.
  12. Cast mainstream “AI” related research in TCS in terms of the social democrat’s alignment problem. See what is out there at conferences like STOC/FOCS, COLT and NIPS. Check O’Neil’s Weapons of Math Destruction, Dwork et al’s line of work on algorithmic fairness, the work on learning non-discriminatory predictors, etc.

On gatekeepers, hormones and unethical research practices?

Background

Many, though not all, trans people who transition get hormone replacement therapy (HRT).  For transmasculine people, testosterone. For transfeminine people, estrogen and (before surgery) testosterone blockers. HRT can affect most secondary sex characteristics in a way that makes a person more comfortable with their body and helps them in being seen as their preferred gender.

For trans youth, puberty suppressing medication exists. These medications pause puberty for as long as you use it without other consequences. It is completely reversible: stop taking the medication, and your puberty will resume as if nothing happened. (All medications have side effects, but this stuff is super mild.) Puberty suppression is fantastic because it can prevent breast growth, voice deepening, beard growth and male pattern baldness, which are all expensive and time-consuming to undo after they happened. Puberty blockers are for young people who want to go on hormones but are considered too young for that, or for people that are undecided about starting HRT.

Puberty suppression medication and testosterone blockers are only legally available with a prescription in every (?) country in the world. If you want it, you have to get a medical professional to prescribe them.

Doctors have specialisations and avoid treating patients for things outside their area of expertise. That is why historically and presently you probably have to go to a specialised transgender endocrinologist to get a prescription for hormone suppression meds.

The World Professional Association for Transgender Health (WPATH) publishes a document called the Standards of Care (SoC), in which they give widely followed guidelines for medical professionals on how to best treat trans people. They say that, before medically treating an individual,  they need a diagnosis (3-12 months of counselling) and go through the hazing ritual known as the real-life experience (3+ months, no longer recommended since 2011). And not just for HRT, but also to get these puberty suppression medications. This practice is often called the gatekeeper model, in contrast to the informed consent model in which the doctor is only there to inform you of what the medications do and then gives them to you without requiring you to jump through any hoops.

Many places that treat trans people have waiting lists measured in years.

Imagine you are 14. The first symptoms of puberty start to happen, and you find out you don’t want those. Even if your parents are supportive and you can immediately get a referral from your GP, it might take two or more years before you get puberty blockers. Entirely unnecessarily, you have two years of unwanted puberty happening to you. Everyone involved knows this is unnecessary. Too bad, you’ll have to do voice therapy and laser hair removal to fix it later.

(DIY medication exists, but it is technically illegal, 95% of trans people recommends against it for reasons unknown to me, and treatment providers threaten that they will no longer help you with anything when they find out you self-medicated.)

Section of Doubt

To my best understanding, the above is a more or less accurate description of the state of affairs for many people. I consider it grossly negligent of medical professionals and policymakers to force trans people through unnecessary additional puberty, and of trans advocates that they are not making more of an issue out of this.

I do believe that people, in general, have reasons for doing what they do and believing what they believe. Considering that my view of the situation is very different from most people, I am probably missing something important. I have no idea what it might be or who is right.

Storytime

I was almost 18. I was in the process of convincing my gatekeeper/psychologist to write me a prescription for testosterone blockers. It was not going super well. I am not traditionally feminine. I didn’t desire to wear female clothing. I don’t “feel like a woman.” I wasn’t sure whether surgery would be right for me, nor whether I wanted voice therapy to sound more feminine. I just knew that I did not want any of testosterone’s further effects. I wanted HRT. I needed HRT. Only after that, I would have the time and peace of mind to think about additional steps.

I was unable to articulate the required narrative and too honest to tell the proper lie on 100% of the hundreds of arbitrary, unrelated or stereotyping questions that were supposed to measure transness. What should have been six months of counselling for getting my diagnosis, had already lasted eight months. The end was not in sight. My therapist was not planning to help me soon, not even with puberty blockers.

But then it happened. The clinic was running a scientific study on differences in brain structure between cis and trans people, and whether taking hormones affected that. They needed trans participants aged 14-19, who just started puberty blockers, to do some mental tasks in an MRI scanner, with a follow up a year or so after going on hormones. I was asked to participate in the study. If I said yes, I would get the puberty blockers and my continued masculinization would stop. If I said no, I would not start medication for months. I would probably have started growing facial hair before I would get medical help.

I felt so lucky; I immediately said yes. My body would be mine, without nature forcibly intruding on my happiness. For the first time in years, I did not want to die. I was so grateful to the researchers for letting me participate in the study. They allowed me to get my hormones.

Epilogue

This event happened over five years ago. The details might be off, but I am confident about the general outline. I am still very pleased with having HRT. I changed my name at that time as well. My old name was ugly; my new name is the best name. These days I am consistently gendered female. I’m happy about being gendered female; it saves me from a lot of transphobia and toxic masculinity.

Today, I first thought back to what happened all those years ago. Only now I realise that this story can be accurately reframed as “my doctor withheld me my treatment to force me to participate in a study”, and that ethics boards probably don’t appreciate extortion of patients. On the other hand, ethics boards do approve of the gatekeeper treatment model, so their opinion might not be the best guideline for what is good or bad. I don’t know what to think of this.