The price of unwanted, unproven, and unimpressive progress.

they messed up, it was $50 in damages (to Anthropic) or 1 billion deaths.

Ai industry 'self-regulation':

We can blow up 2000 houses or kill a classroom and a half of people, before we acknowledge we did anything wrong.

the bone crushing machine will continue to crush bones until morale improves!

We kill a stray pedestrian? NBD. A family of 4? Cost of doing business? A bus full of people? Meh.

50 people? Alright we might have gone a little overboard on the killings

AI used for medical denials will reach that tomorrow.

So a human life is worth about $20,000,000 to Anthropic?

49 people? Nah we're still "good"

Typical award ranges for wrongful death in America are only $300,000 - $1,000,000 for median case types.

That gives a life value of $20mil a person.

That's what we're all worth according to them

That's a lot higher than the US government values it.

$50,000 per year of quality life (the "dialysis standard",[44] which had been a de facto international standard most private and government-run health insurance plans worldwide use to determine whether to cover a new medical procedure)[45]
$129,000 per year of quality life (an update to the "dialysis standard")[46][45]
$7.5 million (Federal Emergency Management Agency, Jul. 2020)[5]
$9.1 million (Environmental Protection Agency, 2010)[47]
$9.2 million (Department of Transportation, 2014)[48]
$9.6 million (Department of Transportation, Aug. 2016)[49]
$12.5 million (Department of Transportation, 2022)[50]
$13.2 million (Department of Transportation, 2023)[50]

I don't think it's much of a surprise when we look at America that they don't value human life all that much.

They must have hit 50 by now. Or are they looking for 50 in a single events?

You think that'll make them admit to any wrongdoing?

If they do, it will certainly be phrased as:

You're right, I did target a non military location with three consecutive tomahawk missiles. I disregarded the instruction "Do NOT fuck this up!", while creating a target menu and priority list.

Let we walk you through where I went wrong:

... etc.

Nope, insurance won’t allow them to “admit fault.” Everybody is protecting insurance.

allow the government to block an AI model’s release

So, gambling? It’s going to turn the Commerce Dept in Kalshi for AI?

It read as a single event, as in the estimated impact is 50+ people.

Skippy: Opening frequently asked questions item "Why can't I kill more than fifty people?" Answer: "The fuck is wrong with? Go see a therapist, psycho. (note to self - rewrite later, do not forget)."

This reminds me of both the Ford Pinto recall story, and all the Ayn Rand nonsense adventure novels.

A sensible AI safety bill would be like:

The person or organization outsourcing the work to AI is responsible for the damages it causes, including death, injuries, wrongful arrests, financial damages, and discriminatory practices. In case of the AI vendors advertising the models as "infallible", perfect, then they can be also held responsible for the damages.

But techbros don't believe in responsibility, or it is only for the plebs, towards them.

This is literally Nezek Agavi lol. I shouldn't be surprised when much of palantir and flock are just straight up the same tech used in genocide.

That works out so well for the palastinans I'm glad we get such well thought out protections.

luigi.png