55 comments

[ 3.3 ms ] story [ 92.5 ms ] thread
But then what? Ditch the people to avoid responsibility? Jail them to prevent suicide? (and with that, avoid responsibility) Or actually get to the root cause and fix that, even if the root cause is society in one way or another...?
Ideally you use the predictive technology to enhance and check the work of humans. In reality, I agree that people usually just get ditched in favor of the technology that works most of the time...but not all (self-driving comes to mind).
Military readiness is organizing and preparing the military for success in the field. If members are unwell, then they need to be identified, and then treated.

Suicides are a big problem affecting unit readiness at the moment, as depression not only lowers the effectiveness of individual people, but the suicide also affects unit morale.

Not everyone is comfortable with seeking help, especially in the armed forces (due to a variety of reasons such as culture), so if there are more tools to visibly see these issues (like we would with standard medical issues), then that makes it much easier to open up dialogue with our soldiers, and begin treatment.

But if the culture makes it hard, technology is hardly going to fix that, right? Culture is a human problem, not a technology problem. Technology is just a tool, and perhaps a force multiplier, but a change in culture?
Correct; but that doesn’t mean there isn’t any utility in new technology for improving our culture so long as the people using it want to do so.
That's where the real fun of it lies. Imagine you have at your disposal every tool to modify individual behavior, but no tools to modify social systems? What second-order effects result from solely relying on individual interventions? If there are ways that make people feel ok with the soul-crushing weight to social ills, is it right to employ them?
> Not everyone is comfortable with seeking help, especially in the armed forces (due to a variety of reasons such as culture), so if there are more tools to visibly see these issues (like we would with standard medical issues), then that makes it much easier to open up dialogue with our soldiers, and begin treatment.

And once again, we're seeing technology being attempted to use as a band-aid instead of fixing the actual problems - the culture of toxic masculinity that is not just in the mindset of the soldiers but also in almost all advertisements, the fact that many seek to join the military not out of their own will (e.g. to defend the homeland) but to escape poverty, or the fact that many get sent to the military by their parents either because of "tradition" or because they are seen as "rebellious" and the military should "fix them".

Humans aren't cogs in a machine, and projects such as this are only one thing: another layer to make sure the cogs perform as required. Dehumanizing.

So, I disagree that this is a band-aid, because you can be depressed without realizing it. War is brutal, and people are subjected to a wide range of emotions that can manifest consciously or subconsciously, so some people in-tune with their thoughts may have issues without realizing it.

However, with regards to the rest of your comment, there is a population that sees toxic masculinity as just traits of normal masculinity, with all of the problems that comes with. The fixing of actual ingrained cultural issues is a culture war that only recently picked up, and is not going to resolve anytime soon.

Meanwhile, there are people that are suffering who do not understand that their (to them) desirable identity traits are causing them harm, and potentially do not understand that they are going though depression. I do agree that it is depressing that soldiers are treated as fungible, but mental health is probably the only place where you're going to see humanity injected into the armed forces.

I can imagine the dystopian sci-fi work now, of someone who has been tagged in the governmental credit check system as being 'at risk of suicide'. They are systematically denied access, othered, and subtly offered help for mental health problems (which they did not perceive themselves as having previously before being suggested) causing them to be pushed to the edge of sanity.
worse yet - that individual has incurred healthcare costs at a greater rate than their "peer group" and the nudges turn away from mental health assistance to subtle suicide encouragement for the "common good" (lowering aggregate healthcare costs).
Or you find yourself repeatedly thrown in the looney bin for a 72 hour suicide watch. And sent the bill (see Baker Act and other constitutionally-dubious state laws).
The inability to hold down a job is a very common symptom of mental illness. In other words, it is no secret that your mental health hinges on whether or not you contribute to the economy.
I think you may have causality backwards there. Its more likely that people with mental health issues have problems functioning in the usual work environment, they don't mesh and have trouble keeping the job because of it.
Well I do fully believe that for most people their sense of self worth hinges on having responsibilities and a regular schedule, but that can mean being a mother, volunteering on a regular basis, being a part of your retirement community's bingo club, or whatever else.

But what is also true is that losing your job can be a precipitous plunge into mental illness, as your outside world that was regularly structured turns to chaos suddenly. And of course, if your external world becomes chaotic, your internal world is liable to be chaotic as well--'as above, so below' and all that.

As along as identifying them as at-risk doesn't equate to labeling/condemning them as at-risk or suicidal without protections and more traditional steps/measures (not that those are fool proof either).
As likely to be used for suicide prevention in the end as older drone projects were for "delivering aid during disaster scenarios" with the required capability to identify individuals. I would say the road to hell is paved with good intentions, but it turns out giving research some transparent lie to tell themselves is more then enough.
Maybe not for directly delivering aid, but drones have been pretty successful in being useful for search/rescue/mapping disaster areas.

Sometimes technology isn't immediately useful in the exact way we think it will be.

Sent from my DARPA nuclear-war tolerant communication terminal.

You mean like using the tool to systematically identify the small changes in suicide likelihood due to parameter variation of a subject's interactions/environment and then maximizing the likelihood? A suicide inductor.
For example. Or more likely just not just searching for suicide indicators

>but, rather, on aggregating preconscious brain signals to determine what someone believes to be true. The screening process envisioned could involve presenting various statement stimuli of behavioral health relevance, such as biographical information, actions, or intentions (e.g., I want to end my life/enjoy my life) to measure preconscious responses. NEAT would triangulate responses to aggregate evidence and determine if the person reading the stimuli statements believes they are true, false, or indeterminate.

Wanna bet that it will very quickly be used for detecting other things the governments deems "unwell". Resulting in another kind of "help". The Simpsons clip sprang to mind https://www.youtube.com/watch?v=zO0-mofj9EY&t=3s

It is yet another targeting algorithm that has suicide prevent as one very easily replaceable scenario.

https://knowledge.insead.edu/blog/insead-blog/africas-drone-...

>”In Africa, doctors are also placing orders remotely, but for a rather different kind of sustenance: blood.

> “The plasma is delivered by drones to far-flung communities in Rwanda and Ghana within minutes of the order being placed by text message, often in a context of life-and-death like childbirth gone wrong. They are the world’s first national-scale solution for the healthcare and emergency needs of people living in places not served or hard to reach by land transport. What started with blood supplies and vaccines now also includes medicine, contraceptives and personal protective equipment (PPE).

ZipLine have been doing this since 2016, and are now doing it in the USA too.

I think this is a valuable contribution to this conversation. You have shown how drones are definitely being used in critical situations. I can't see why you're being downvoted. I always try to support comments that are being downvoted for the wrong reasons (in terms of the HN rules).
Didnt downvote, but you are missing the point. The bad part isnt the drone but the targeting algorithm. See stuff like https://www.darpa.mil/news-events/2017-06-28

>Under the FLA program, the only human input required is the target or objective for the UAV to search for—which could be in the form of a digital photograph uploaded to the onboard computer before flight—as well as the estimated direction and distance to the target.

Where are you quoting from? Why would DARPA have any need to conceal that it's research had a military purpose?

I can't find anything that says drone projects were originally pitched for aid delivery.

One example were those self destructing drones https://www.zdnet.com/article/darpa-develops-disappearing-de...

> Announced this month, the US research agency said Project Icarus will create a fleet of tiny, single-use drones which will deliver emergency supplies -- such as food and medicine -- to remote areas during epidemics or disasters.

> However, once their mission is over, each drone will "vanish" after landing thanks to the use of special materials which can transform their state or shatter into mere particles.

I however remembered it more in the context of this one https://www.darpa.mil/news-events/2017-06-28

>Under the FLA program, the only human input required is the target or objective for the UAV to search for—which could be in the form of a digital photograph uploaded to the onboard computer before flight—as well as the estimated direction and distance to the target.

Neither autonomous targeting of specific individuals nor the capability to "vanish" are something that has much application in disaster relief. Its the evolution of slaughterbots.

"NEAT would identify psychological and behavioral changes before they impact military readiness."

I think the idea is to identify soldiers and other personnel who are at risk of becoming suicidal during a military operation. This seems like a good idea to me, given the extent that the safety of soldiers depends on the mental acuity of their colleagues.

On the other hand, if I had this technology and was sufficiently antisocial, I could use it to keep an 'interrogation subject' right on the edge of killing themselves for an extended period of time. I could use it to recruit soldiers for fatalistic operations. I could use it to kick out members of any operation whose moral compunction was detected to be causing them too much distress.
Yes, it's possible to find appalling uses of most things.
Or it could be a way for them to fire suspicious, defiant, or personell they suspect are whistleblowers, without any questions asked.
They are >aggregating preconscious brain signals to determine what someone believes to be true.

Its a terrorism divining rod. That needs zero change to be used for detecting simple opposition as well.

Now this cuts out the annoying governmental process of having to find your future suicide bombers. You can just look them up in a database and assign an agent immediately!
This 2022 remake of Minority Report is really a snooze.
If I'm understanding this correctly, it essentially tries to tell if someone agrees with a statement ("I want to end my life") by looking at "preconscious" brain signals, regardless of what the interviewee answers consciously. So is it like an MRI for lie detectors? I wonder how it compares to the faults of traditional lie detectors.

Besides the weird "precog" minority report vibes, it makes for an interesting validation problem.

From the brief overview provided by DARPA it's not entirely clear what methods or system is envisioned to do the "preconscious" screening. Presumably some kind of measurement of brain waves, but that seems way ahead of where the science is at.

In any case I wonder if this approach is really the right target. It's true, the suicide rate in military (active duty and veteran) is tragic, but also elevated in the general population. I'd venture a guess that the issue with suicide reflects a tremendous stigma attached to behavioral disorders such that sufferers are dissuaded from seeking evaluation and treatment until conditions reach disabling proportions or maybe not even then.

I have serious doubt that technological methodologies alone will be an adequate solution. Of course it's possible that new tools will move the ball downfield and play a role in extinguishing long-standing prejudices but it remains to be seen if that will happen.

I hate to be a cynic, but my feeling is that it's more applicable to lie detection and the suicide prevention is a good application (but probably far from the main goal). Cut out all the suicide-specific wording, and it sounds like generic lie detection that military or law enforcement would have a lot of interest in.

"The screening process envisioned could involve presenting various statement stimuli...or intentions (e.g., I want to [do this bad thing]) to measure preconscious responses. "

My personal opinion (backed up with a certain amount of experience) is that much of these suicides are due to a soldier being traumatized, then coming home to try to cope with trauma responses. Some then turn to drowning the trauma with alcohol and other bad stuff. That's when the spiral really accelerates.

Unfortunately the stigma you mentioned feels so much worse when someone also hates themselves. They don't recognize the stigma for what it is and start believing they don't deserve to get better.

And yes this seems to be a huge problem across humanity. Overdoses and suicides are at historically high levels. And celebrities with tremendous resources still suffer the same way. Which tells me the technology just isn't there.

No question military experience can be extremely traumatic even without exposure to combat. People vary greatly in "resilience", some are very sensitive to trauma but may have no way to know that about themselves before encountering the damaging situations.

Thing is mental health conditions in general are a huge problem: "Neuropsychiatric disorders are the leading cause of disability in the United States, accounting for 18.7% of all years of life lost to disability and premature mortality." [0] The numbers vary in different countries/regions but the prevalence of such disorders is still non-trivial.

In addition most people don't seek or get adequate treatment, I think stigma is a big factor contributing to making a bad situation worse.

The elements you talk about are certainly true. It really would be a major change for the better if technological advances could make a difference. Given the developments of the last few decades I wouldn't say it can't happen. IMO the jury's out on the question, conceivably research will eventually have something to offer.

[0] https://www.healthypeople.gov/2020/topics-objectives/topic/m...

Ugh.. that's none of their damn business? If a person needs help they can reach out and you can make remove any barriers preventing them but having your mental disorder identified against your will sounds horrific. Ends don't justify means!
Where does it say it will be against someone’s will?
> ...since patients will often tell their clinicians what they think the clinician wants to hear rather than how they are truly feeling

Right in the article. The idea is to find out medical information patients aren't willing to share with their physician.

> The idea is to find out medical information patients aren't willing to share with their physician.

Plenty of routine medical tests already do that.

*sigh* HN comments are entirely predictable.

Tools that help understand and detect suicidal behaviour could also help refine an understanding that people can use to spot potentially pre-suicidal behaviour in the people they love.

You know that suicidal people often don't, can't or won't reach out for help at all, right? And may in the days before their suicide appear to be dramatically better, serene, even peaceful?

There are likely other common patterns of pre-suicidal behaviour (unusual secrecy for example), and a tool that can help clarify what the reliable indicators are, really might stop people killing themselves. We want this, don't we?

In DARPA's case, if those people are in the service, it is their damn business; they have a duty of care.

Away from DARPA, better processes might have stopped a horrific airline disaster in recent years. There are some people in highly significant, routinely stressful roles who are regularly psych-screened anyway. Those screenings need better tools to be more useful.

> that people can use to spot potentially pre-suicidal behaviour in the people they love.

The people they love did not obtain consent from them.

> You know that suicidal people often don't, can't or won't reach out for help at all, right?

Yes. First hand. Suicidal and/or mentally ill people have rights such as one for privacy just like you.

> and a tool that can help clarify what the reliable indicators are, really might stop people killing themselves. We want this, don't we?

We don't. Assholes who think using other peoples' death and illness to perform non-consensual medical examination is good do. By this logic, you will eventually even say people's eating habits and behavior should be monitored before they kill themselves by eating food that causes obesity and diabetes before dood addiction starts.

I concur about the predictability of HN comments. When euthanasia pods became a thing in sweden you all lauded it because someone is profiting. Here too, someone will do a startup out of this so it is a cause for celebration. Never mind privacy, free will and self-determination.

> t is their damn business; they have a duty of care.

So things made by darpa never make it to public use and service members have no right to medical privacy like us HIPPA enjoying civilians? Okie...

To your last paragraph: those people agree to consensual and explicitly stated evaluations not non-consensual psychological examination of a medical condition.

We don't want to help people not kill themselves?

This is depressive realism taken to a frightening new level.

What DARPA is talking about need not be incompatible with some relaxation around medically-assisted suicide.

But people kill themselves for all sorts of terrible non-terminal reasons. Teenagers, especially. We should try to help them not do it. And research like this might shed more light.

> By this logic, you will eventually even say people's eating habits and behavior should be monitored before they kill themselves by eating food that causes obesity and diabetes before dood addiction starts.

"By this logic" === "Here's a straw man argument"

> When euthanasia pods became a thing in sweden you all lauded it because someone is profiting.

I am not in that particular "you all", thank you.

> To your last paragraph: those people agree to consensual and explicitly stated evaluations not non-consensual psychological examination of a medical condition.

Yes. But as I understand it, soldiers sign up for a lot of evaluations and even some interventions over which they have no choice.

> We don't want to help people not kill themselves?

It's not help if you are violating consent of a person that has the ability to give consent. Minors and intoxicated people are some examples of people that can't gove consent in which case their legal guardian must give consent.

The shitty thing about the tech scene is the utter disregard for consent because you know better.

Few things I hold higher contempt for over people that violate consent because they know what is best for someone else over that person themselves.

> But people kill themselves for all sorts of terrible non-terminal reasons. Teenagers, especially. We should try to help them not do it. And research like this might shed more light.

Yes they do and they should be helped but only if they or their legal guardian provides consent and is willing to be helped. Have you actually tried to help a person in this situation? Nothing more humiliating or pushing over the edge than your private mental struggles being exposed and well meaning assholes now made your contemplations into an issue you must act on (as in kill yourself or chicken out). Ffs! It's slavery not help what you have in mind. Your intentions don't change that fact.

> ...Here's a straw man argument"

I can see why you would say that but both in law enforcement and tech there are plenty examples of things being done for someones best interest that end up having horrific consequences. 72 hour mental health lockup suicides is one. You lock them up for monitoring their mental health for their own good except now they are so scarred of being treated like psychiatric patient they can't recover from it. Or drunks being locked up so they don't hurt themselves except other prisoners have fun with them.

> Yes. But as I understand it, soldiers sign up for a lot of evaluations and even some interventions over which they have no choice.

Right and if the purpose is to save the government money on dead soldiers it should be stated as such. When framed as a mental health tool, it is only natural to apply it to civilians. Also, Soldiers should also be allowed a chance to consent explicitly just like with everything else.

> It's not help if you are violating consent of a person that has the ability to give consent. Minors and intoxicated people are some examples of people that can't gove consent in which case their legal guardian must give consent.

Nothing about anyone seeking to help a loved one not kill themselves inherently requires consent.

I said nothing about invasive processes, outside of the DARPA situation where they almost certainly have the consent they need already.

> The shitty thing about the tech scene is the utter disregard for consent because you know better.

This is projection.

> Have you actually tried to help a person in this situation?

Yes. And they're still around and living their life decades later. Because they were suicidal about something that could be substantially unburdened if they would just talk. And they weren't going to talk without an intervention (which was staged initially online, as it happens).

> Nothing more humiliating or pushing over the edge than your private mental struggles being exposed and well meaning assholes now made your contemplations into an issue you must act on (as in kill yourself or chicken out). Ffs! It's slavery not help what you have in mind. Your intentions don't change that fact.

Suicide is not a good alternative to humiliation.

If this is where you are right now, talk to someone. I'm not going to continue this discussion because of what you say in the last paragraph I quoted. But if you need help, consider asking for it.

You changed the subject and made it about me. This isn't personal it is a matter if principle. Everyone that can give consent must do so before you perform medical or behavioral analysis if the person. No exceptions. That was my point initially, using ends and emotions to get around this is absolute b.s.
> You changed the subject and made it about me.

I did. Out of concern, as I will whenever I hear people talk about suicide in quite a fatalistic manner. A stranger did this for me only two years ago and I am happier because of it.

DARPA (war department) works on life-saving applications of AI, while private social media works on addictive engagement algorithms to make people miserable and enslave them to dopamine loops.

This strike anyone as a bit off?

"If it saves even one life, it will be worth it!"

Even if it means preemptive curtailment of their liberty, privacy, and choice, whether or not they would actually have committed suicide if left unmolested.

This is such a moonshot. Right now, give all warning signs we know if, if you could identified the 100 most likely people in America to attempt suicide in the next 24 hours that haven’t had a previous suicide attempt or previously announced intentions of self harming you might successfully identify 5 people. Answering the question “who will attempt suicide” is a lot like “who will have a fatal car accident “. There’s a handful of obvious warning signs (drunk drivers), but even those aren’t strong enough to allow any certainty.
Suicidality is 47% heritable. So just get the blood of people who have committed suicide & do some GWAS studies. Then you can look for predictive genetic markers in 23andme data.
A society that criminalizes suicide is a society that keeps its citizens as slaves.