Stories on Dismissed Alarms (and normalcy bias)

[vale verge]

I'm collecting (both true and fictional) stories about warnings that were dismissed, and disasters that happened as a result. I've been using some of these in presentations when I talk about AI risk.

I'm considering bundling them on the website. If I won't do that, I think it's still useful to have a list of these just to give as examples.

Stories

• The Challenger space shuttle blew up because of a faulty o-ring. The day before launch, multiple engineers warned that this would happen.
• The titanic sunk the day that the ship inspector noticed there were not enough life boats
• At 9/11, people took an average of 6 minutes to take action. Thousands took the time to save files / send emails / shut down their computers before evacuating.
• During hurricanes, people check an average of 4 sources before taking action.

Science / psychology

• Normalcy bias. A cognitive bias causing individuals to disbelieve or minimize threat warnings, underestimating both the likelihood of a disaster and its potential adverse effects. Around 70% people are in denial (nomalcy bias), 15% break down, and 15% act effectively.

How you can help

• Share a story!

[severe pilot]

Dismissed Alarms would be a better name

Alarmism is excessive or exaggerated alarm of a real or imagined threat.

[vale verge]

Stories on Dismissed Alarms (and normalcy bias)

[vale verge]

good point, edited

[elfin lodge]

Cassandra from Greek mythology and every monster movie where the adults don't believe the plucky kids who then have to save the day

[smoky glen]

70% link is dead, is this the right one? https://gizmodo.com/the-frozen-calm-of-normalcy-bias-486764924#:~:text=When disaster strikes%2C some people,up until it's too late.

[slate mauve]

I've seen many people in my apartment building not leave during a building fire alarm. While modern apartments are pretty inflammable, I suspect there'd be some kind of study about this.

[bleak hamlet]

I have personal experience with “The Bystander Effect” with regard to the public’s perception and intervention preventing crime from occurring in the moment.

11 years ago, in 2013 when I was 22, I got off work late one night and went downtown San Francisco to meet up with friends around the intersection of Church/Market for Pride Festival. It’s an absolutely massive event in San Francisco, with tens of thousands of people partying in the streets and bars all over town. That specific area (as well as the Castro) are especially crowded at nighttime during this period.

Anyway, I show up, completely sober after work, and immediately meet up with my friends. We’re walking down the absolutely packed street when I see 5 guys and 1 girl (6 total individuals) jumping a single other woman and stealing her purse. The street is packed and absolutely nobody is doing anything other than watching, as they expect “someone else” to do something. This is actively occurring and nobody is doing anything, so after a minute or so of this bizarre and cowardly dance, I take it upon myself to jump into the fray. I step towards the group and stupidly yell “YO,” while winding up a punch on one of them. I immediately get jumped and the crowd now watches me get the shit kicked out of me with the same apathy they felt towards the previous woman, although she was able to get away with her purse during this process.

The point I’m trying to make here is that nobody wants to be the person to break up a fight, to break a major whistleblower story, or to personally put themselves in harm’s way when “there are so many other people that will surely jump in before we need to ourselves.”

This effect is something I see often with AI technology these days. Everyone says “well, someone else will handle it.” The problem is, someone else isn’t.

https://en.m.wikipedia.org/wiki/Bystander_effect

https://www.sfgate.com/crime/article/String-of-robberies-assaults-after-Pride-4648761.php

[bleak hamlet]

I got my nose broken and went to a house party later that night irregardless. I had it fixed the next morning, went into work with a cast on my nose and got a raise on the spot after telling my manager the story. Everyone praised this heroism after the fact, but nobody wanted to be that person in the moment. This includes the police who I filed a report with the next day, and who only managed to arrest one of the individuals involved when everything was said and done.

The point I’m trying to make here is that human beings believe “the group won’t allow something like that to happen,” but in larger numbers, the amount of people involved with any single issue actually prevents any specific individual from feeling the need to intervene personally.

The bystander effect needs to be properly understood with regard to AI-legislation and whistleblowing. That’s my point here. “If we don’t do anything, who will? needs to become the new mindset, because the old one is clearly ineffective.

[severe pilot]

Thanks for sharing and trying to help

I think is pretty much the same as https://en.wikipedia.org/wiki/Somebody_else's_problem

[elfin lodge]

That sort of engagement is the difference between subjects and citizens

[severe pilot]

How do you envision this page? I think only text would be bad. Maybe we could hire a web developer to do some type of horizontal or vertical timeline that can be interacted with

@vale verge

[severe pilot]

and so people can even filter for type of danger

checking all the #1172181242339409961 and AISMemes should give us lots of examples

• https://incidentdatabase.ai/

[wild cargo]

@small glade maybe what you did can be used to help this?

[small glade]

I'm sorry, im not sure what you're referring to?

[wild cargo]

The ignord warning shots you gave me

[small glade]

oh right! Can i just send these here?

[wild cargo]

You should!