Dawn Drescher

Karma: 466

There are critical gaps in the accessibility and affordability of mental health services worldwide: In some countries, you have to wait for years for therapy, in others you get at max one session per month covered by the health insurance, in others therapies for particular conditions like cluster B personality disorders are virtually nonexistent.

We want to leverage LLMs to fill these gaps and complement regular therapy. Our product is in development. We’ve based it on Gemini and want it to interface with widely used messaging apps, so users can interact with it like they would with a friend or coach.

I’ve previously founded or worked for several charities and spent a few years in earning to give for work on invertebrate welfare and s-risks from AI.

You can get up to speed on my thinking at Impartial Priorities.

AI Mental Health Chatbots for Low-Resource Settings: A Prioritization Framework

Dawn Drescher1 Dec 2025 17:41 UTC

6 points

0 comments16 min readLW link

Dawn Drescher 25 Oct 2025 18:21 UTC
3 points
1
in reply to: Prometheus’s comment on: Prometheus’s Shortform
[Same as what I replied to your DM:] Yeah, exactly! Systems like that were on my mind a lot around 2021–23 when we launched Impact Markets, which is now GiveWiki. You could do the payouts if humanity has survived for another year, has survived AGI, has survived a year with AGI, etc. You could also chain retrofunders with different time horizons. We tried to get something like this off the ground for about 2 years, but couldn’t find any retrofunders anymore after FTX collapsed. (Some of the regrantors were interested.)

https://impactmarkets.substack.com/p/chaining-retroactive-funders

If you can find a retrofunder for a system like that, you can make it happen!

Dawn Drescher 23 Oct 2025 21:50 UTC
3 points
0
on: [Intuitive self-models] 7. Hearing Voices, and Other Hallucinations
New study: Corollary Discharge Dysfunction to Inner Speech and its Relationship to Auditory Verbal Hallucinations in Patients with Schizophrenia Spectrum Disorders.
Background and Hypothesis
Auditory-verbal hallucinations (AVH)—the experience of hearing voices in the absence of auditory stimulation—are a cardinal psychotic feature of schizophrenia-spectrum disorders. It has long been suggested that some AVH may reflect the misperception of inner speech as external voices due to a failure of corollary-discharge-related mechanisms. We aimed to test this hypothesis with an electrophysiological marker of inner speech.
Study Design
Participants produced an inner syllable at a precisely specified time, when an audible syllable was concurrently presented. The inner syllable either matched or mismatched the content of the audible syllable. In the passive condition, participants did not produce an inner syllable. We compared the amplitude of the N1, P2, and P3-components of the auditory-evoked potential between: (1) schizophrenia-spectrum patients with current AVH (SZAVH+, n = 55), (2) schizophrenia-spectrum patients without current AVH (SZAVH−, n = 44), (3) healthy controls (HC, n = 43).
Study Results
The HC group showed reduced N1-amplitude in the Match condition (relative to Passive and Mismatch), replicating our previous results. In contrast, the SZAVH+ group showed the opposite effect: enhanced N1-amplitude in the Match condition (relative to Passive and Mismatch). The SZAVH− group showed reductions in the Mismatch condition (relative to Passive and Match).
Conclusions
This study provides empirical support for the theory that AVH are related to abnormalities in the normative suppressive mechanisms associated with inner speech. This phenomenon of “inner speaking-induced suppression” may have utility as a biomarker for schizophrenia-spectrum disorders generally, and may index a tendency for AVH specifically at more extreme levels of abnormality.

Dawn Drescher 23 Oct 2025 21:40 UTC
3 points
0
in reply to: Viliam’s comment on: Narcissism, Echoism, and Sovereignism: A 4-D Model of Personality
Yeah, makes sense. Something else I failed to mention is that pathology also requires that we’re not simply dealing with a reasoned decision of someone who could’ve just as soon decided something else, but with a decision that is so multiply overdetermined by traumatic adaptations that it’s almost impossible for the person to do anything else. So the type of decision process also makes a difference.

Dawn Drescher 20 Oct 2025 14:24 UTC
3 points
0
in reply to: Viliam’s comment on: Narcissism, Echoism, and Sovereignism: A 4-D Model of Personality
Yeah, I think I agree, and I have hunch as to why. Let me know what you think.
I think an objection such as that they can’t help people if they’re dead misses the point that working themselves to death might just be constructed to be the right thing to do according to utilitarian values, e.g., because it’s the crunch time in their space.
What’s going on is rather that all of these behaviors are adaptations to terrible environments, and people can only really heal once they are out of the environment that makes the behaviors adaptive, that requires them. Otherwise “healing” would be just doubling down on the same adaptations while learning more mature defenses and reality testing. But the world is one ongoing catastrophe, so there’s no getting out of that one.
So to find out whether the behavior is pathological, we’d have to find out whether the person actually has the sort of utilitarian values that make it consistent or whether they’re acting from a place of, e.g., helplessness.
How I would phrase this item is “can’t accept helplessness,” because I imagine that in many cases these people are not actually such consistent utilitarians but that they feel helpless in view of all the suffering in the world. But because helplessness is such an intolerable feeling, they prefer to feel responsible for all the suffering and hence guilty of it, and then this guilt or responsibility propels them into the catharsis of self-destruction.

Dawn Drescher 16 Oct 2025 18:05 UTC
1 point
0
in reply to: Gunnar_Zarncke’s comment on: [Intuitive self-models] 3. The Homunculus
Gosh, yeah… Is that what’s called the watcher? Can I even literally watch this process unfold or is it by necessity what I’m doing now, looking at my memories and trying to collect and timeline everything that I remember?
If D is the decision process, then perhaps during the process there can only be D whereas afterwards I can have S(D) and thus become aware of D? I feel like I’m always looking at an “echo” of the real thing when I’m writing comments like the above.

Dawn Drescher 16 Oct 2025 17:55 UTC
1 point
0
in reply to: Viliam’s comment on: Narcissism, Echoism, and Sovereignism: A 4-D Model of Personality
Do you just mean that there are some dimensions along which it’s unique among the items in the list (each of them is probably unique along some dimensions), or do you mean that it’s not pathological and so shouldn’t be on the list?

Dawn Drescher 14 Oct 2025 17:44 UTC
3 points
0
in reply to: Steven Byrnes’s comment on: [Intuitive self-models] 3. The Homunculus
Ty! Yeah, I think I’m trying to figure out whether I have such a noun and if so if it’s the same as the self, so the question you don’t want to weigh in on.
I observe processes in me that take the shape of:
1. Hypotheticals bubble up and play out in a few different ways:
  1. I’d conventionally describe this as “I’m considering alternatives,” but really the ideas come from somewhere uncontrollably.
  2. Something evaluates the desirability of these alternatives in a chill, nonjudgmental fashion and ranks them according to how in line with my values and goals they are.
  3. There is also this part that has become much more chill over the past year that threatens to self-shame and self-punish me if I end up executing on one further down the list.
  4. I execute one the top option, which I would conventionally describe as “I decide what to do,” but I’m not sure if there’s a decision involved. If I make a spreadsheet and discuss it with people and then execute the top option, it feels like a decision process, but if it’s a decision whether to ask someone one question or another question, the outcomes seems to flow from the ranking without the intervention of a noun.
2. Actions just happen because they were not tagged as being potentially controversial:
  1. I observe myself picking up roasted peanuts, which feels ego-syntonic, or
  2. I observe myself making an annoyed sound, which I regret, which feels ego-dystonic and was previously followed by self-shaming.
Insofar as these evaluations, the chill one and the threatening one, have to do with me, this body/algorithm/identity, I’ve been referring to them as my self. I think that is already a more narrow definition than the one that is widely used that I should switch to once I get a fuller picture of everything else that is also self.
When I look for the homunculus, I look at the moment between ranking and execution. But in most everyday situations, there’s nothing there.
Willpower also doesn’t map clearly to anything. If I want to finish part of a software project some evening despite being tired because someone else depends on me, I have this part that tells me that I should finish it because someone depends on me, which becomes part of the ranking procedure. Previously the threat that I’d self-punish if I don’t deliver was also part of the ranking procedure.
It’s hard to tell for me whether my implicit model of a door knob is that it’s hard to turn rather than that I have trouble turning it. Maybe? When it comes to taste, “X tastes bad” always (or as far as I can remember) seemed like a linguistic shortcut to me rather than a meaningful statement about the external world. Accepting moral antirealism, i.e. also seeing “X is morally bad” as a linguistic shortcut, is something that I only became convinced of 10 years ago when a friend of mine got me to consider it seriously for the first time.
If someone were to tell me, “You owe me!” and it’s plausible, I think that would cause some kind of stir in me that a statement like, “You’re toxic!” doesn’t cause anymore. The first seems to still connect to something that is clearly not viridical but that I’m still reifying whereas I don’t seem to believe in the second anymore.
So idk, I seem to be in some kind of messy state where it’s super hard for me to find this homunculus, and where I and my self are fairly distinct to me but not fully, still blur into each other in some contexts.
Now I don’t doubt that other people have a clear homunculus like that. I’m often puzzled by the reluctance some people display to accept that we might be in a simulation or that copies of them would feel the same as they do or that AIs can be conscious in the same sense they are. So I am a bit weird (though not by LW standards), but figuring out exactly in what ways my perception is different from the conventional one eludes me.
Oh, until 2013 I had a process that narrated all my decisions. The ranking and execution happened as always, but there was this separate process that observed the ranking and execution and, by trial and error, tried to construct narratives of why the execution was the one that it was. There usually made some sense, but in extreme situations they were also often clearly self-deceptive. I stopped doing that in late August 2013. Maybe that was a kind of homunculus illusion?
I’m almost done reading your sequence, and I looove it! Lots of awesome insights! Especially the application to BPD (and by extension other PDs) is very interesting to me!

Narcissism, Echoism, and Sovereignism: A 4-D Model of Personality

Dawn Drescher14 Oct 2025 11:18 UTC

7 points

6 comments14 min readLW link

(impartial-priorities.org)

The Narcissistic Spectrum

Dawn Drescher12 Oct 2025 15:46 UTC

32 points

0 comments22 min readLW link

(impartial-priorities.org)

Dawn Drescher 12 Oct 2025 14:31 UTC
3 points
2
in reply to: Linda Linsefors’s comment on: [Intuitive self-models] 3. The Homunculus
I have the same problem. I can’t find anything that might correspond to this homunculus, and I remember befuddlement when I first discovered that some people find it unintuitive that there could be multiple digital copies of themselves. But I’m not in a no-self state.

The Architecture of the Narcissistic False Self

Dawn Drescher8 Oct 2025 17:39 UTC

4 points

0 comments12 min readLW link

(impartial-priorities.org)

Dawn Drescher 6 Oct 2025 12:52 UTC
1 point
0
in reply to: Gurkenglas’s comment on: The Sadism Spectrum and How to Access It
That’s another interesting hypothesis!

The Sadism Spectrum and How to Access It

Dawn Drescher5 Oct 2025 23:09 UTC

14 points

2 comments20 min readLW link

(impartial-priorities.org)

Beyond Control: The Strategic Case for AI Rights

Dawn Drescher12 Aug 2025 14:05 UTC

−10 points

1 comment3 min readLW link

(impartial-priorities.org)

Dawn Drescher 10 Aug 2025 22:00 UTC
3 points
2
in reply to: Viliam’s comment on: Breaking the Cycle of Trauma and Tyranny: How Psychological Wounds Shape History
In James Fallon’s book (from 2013) it sounds like – though no one really knows – that traits associated with psychopathy (like the so-called “warrior gene”) are likely epigenetic, what I mention above. So preventing wars should gradually get rid of it.
But people with all the genetic predispositions toward psychopathy can still grow up to become perfectly prosocial folks with a good-enough, peaceful, loving upbringing. I know some. No affective empathy, no guilt, etc., but they would be quite disappointed in themselves if they harmed someone, sort of how we’re disappointed if we don’t study and fail an exam.
Then again I imagine that even in a perfectly peaceful society, psychopathy, neurologically/genetically, will continue to exist at some low rate.

Breaking the Cycle of Trauma and Tyranny: How Psychological Wounds Shape History

Dawn Drescher10 Aug 2025 8:46 UTC

46 points

6 comments12 min readLW link

(impartial-priorities.org)

With the Future of the World in Your Hands, Think for 6.77 Years!

Dawn Drescher9 Aug 2025 10:44 UTC

1 point

0 comments10 min readLW link

(impartial-priorities.org)

Dawn Drescher 8 Dec 2023 13:22 UTC
2 points
0
in reply to: William the Kiwi ’s comment on: The GiveWiki’s Top Picks in AI Safety for the Giving Season of 2023
Does this include all donors in the calculation or are there hidden donors?
Donors have a switch in their profiles where they can determine whether they want to be listed or not. The top three in the private, complete listing are Jaan Tallinn, Open Phil, and the late Future Fund, whose public grants I’ve imported. The total ranking lists 92 users.
But I don’t think that’s core to understanding the step down. I’ve gone through the projects around the threshold before I posted my last comment, and I think it’s really the 90% cutoff that causes it. Not a big donor who has donated to the first 22 but not to the rest.
There are plenty of projects in the tail that have also received donations from a single donor with a high score – but more or less only that so that said donor has > 90% influence over the project and will be ignored until more donors register donations to it.
Ok so the support score is influenced non-linearly by donor score.
By the inverse rank in the ranking that is sorted by the score. So the difference between the top top donor and the 2nd top donor is 1 in terms of the influence they have.

Dawn Drescher 8 Dec 2023 13:12 UTC
1 point
0
in reply to: William the Kiwi ’s comment on: The GiveWiki’s Top Picks in AI Safety for the Giving Season of 2023
It displays well for me!