Hacker News Reader

Comment by bananapub | Hacker News Reader

this is important to think about, but more important is to not ever ever let this sort of nonsense stop you blaming a human.

if people outsource "target detection" to some software, they personally are responsible for what it does.

if people outsource the actual murdering to some software, they personally are responsible for those deaths.

a human made the decision to permit this and the entire consequences of it should be on their head. we can never ever let the people who wrote and/or approved the software escape blame for what it does.

we're already seeing this with self-driving cars - which individual at uber should be serving sentences for their cars killing someone? or did we accidentally make murder completely legal as long as you obfuscate the cause chain enough?

View on HN

UHC is planning on releasing AI for their healthcare denials, they’ve been murdering for quite awhile without AI so I doubt this will change anything

Indeed. As we also need to introduce personal responsibility in corporations at large. Corporations = people, but for now, they have a free pass at poisoning the living, killing whoever they need to in places devoided of the rule of law, manipulating politics at large. Heck, some corporations even benefit from in-house police powers and privatized legal systems (when permitted in some places). But that wouldn't be enough. We also need a reformed legal system where crimes are not considered the fault of someone, but more aptly, an evenly shared social failure. Because in fine, we are all responsible for the crimes we allow to be perpetrated. When someone rapes a woman, kills another, or do anything against anyone, we should always question society and ask: how come we the people let that happen? Crime is only an answer to social disengagement, never an isolated incident. Yes, put aside perpetrators until they realize what they've done. But never loose track of the fact that society is sick, and must be healed. Justice as we know it is nothing but an organized revenge system. It will never lead to sanity unless root causes are systematically examined.

What are these dramatic sentences that people are supposedly serving for killing someone? Perhaps we can have Travis or Dara serve the same sentence that someone from their city, SF, who drives drunk across to the other side of the road and then kills a bicyclist: exactly nothing.

Or perhaps we should have them serve the same sentence as someone who takes a turn across a pedestrian walk signal and flattens a toddler and her father. No jail time? Community service?

What’s this idea that this is a crime in any sense? I’m sure Uber can hire someone to do community service per killed pedestrian. That does simplify matters.

I completely agree, if you use software or tech to aid you in killing, you are still fully responsible for the deaths.

Now, let's continue this line of thought. What if people die because your AI told you to not take action when you could have easily intervened to save a life?

That's not a serious issue and, going by US law at least, you're never obligated to intervene to save someone unless you created the situation that threatened them, have a specific special relationship with that person (i.e. parent), or have accepted a role where duty to rescue is part of your obligations (i.e. firefighter).

As an uninvolved third party witnessing some harm you will face no penalty for not intervening.

It is, of course, encouraged to intervene and laws have been written to ensure that a reasonable effort to intervene won't cause you legal issues but you are never obligated to do so.

I'm angling more towards health care and less towards good samaritans.

Oh - well in that case a health insurer has a contract to provide reasonable care... but I think it's pretty clear (especially with UHC's absolutely insane 90% inaccurate rejection bot) that there are bad faith actors involved here.

To be slightly more depressing - before AI auto-rejectors existed companies would just hire doctors and pay them in a way that encourages (I'd hazard to say intentionally) doctors to deny claims as quickly as possible - Cigna was recently caught[1] doing this to an extreme.

There is a deep rot here.

1. https://www.propublica.org/article/cigna-pxdx-medical-health...

Far from being an original line of inquiry, this is covered by the legal concept of Duty to Rescue. Short story is it differs by country, but generally you're not required to save someone's life. In some places you are required to at least contact authorities or emergency services if able.

But that’s not how it works in the real world.

In the real world, “computer said to” provides just enough plausible deniability for _no_ human to be _ever_ blamed by the action the computer decided on.

And this deniability is not side effect its the core reason corporations adopt these technologies for.

/thread

and well said.

Reply on news.ycombinator.com