#Sue the Dutch government

1 messages · Page 1 of 1 (latest)

winged patio
#

Some time ago, there was this project called Urgenda in the Netherlands. They sued the government for not doing enough to mitigate climate risks. The constitutional right to life was used as the main argument. Urgenda won.

We could do the same thing, almost use the same argument.

I'll reach out to some people who were involved in this project.

How you can help

  • research urgenda. Lots of papers, articles, videos, etc were created. What can we learn?
  • help build the argument.

The argument

  • government has duty to care for lives of citizens
  • ai poses catastrophic risks
  • government isn't taking sufficient action

Actions the government should take

  • coordinate a global pause, using asml
  • increase ai safety research budget
  • acknowledge xrisk

Steps taken

  • I've reached out to Mirian Helena Boxberg (who sued Shell, interested in AI according to @Otto Barten) and Dennis van Berkel (urgenda lawyer)
  • Working with Laurens Prins (ex-judge and founder of LegalSafetyLab) who is very much aligned
winged patio
undone igloo
#

Would that work in other countries? Also, I assume the aim is not to win the trial, but to get publicity?

#

and if that is the aim, can we do a quick back of the envelope calculation of the money efficiency of this means of publicity?

winged patio
undone igloo
#

oh waw

winged patio
#

Sent a new mail to dennis:

Beste Dennis,

Hoe is het er mee? Ik zou nog steeds erg graag met je willen praten over het aanklagen van de staat voor het niet voldoende doen om de risico's van AI te adresseren. 

Vorige week heeft het kabinet een visie op AI gedeeld. Fijn dat de overheid iets doet, maar inhoudelijk stelt het bijzonder weinig voor. De grootste risico's van AI (dat werkelijk iedereen dood kan gaan) wordt niet expliciet genoemd, laat staan aangepakt. Van Huffelen herkent "zorgen die genoemd worden in de oproep van techprominenten", maar merkt op: "de oproep richt zich echter in grote mate op toekomstige ontwikkelingen" en dat er nu ook al risico's zijn die "niet opgelost worden met een pauze".

Dat is natuurlijk het hele punt van een pauze ook niet: het gaat niet om de risico's van huidige modellen, maar om de risico's van modellen die er in de toekomst komen. Daarom tekenen 33.000+ mensen de pauze brief, daarom zeggen talloze AI professoren dat het existentiële risico moet worden aangepakt. Daarom hebben de drie meest gepubliceerde AI wetenschappers allemaal hun carrière omgegooid om mensen te waarschuwen. 

Ik kan de visie van ons kabinet niet anders interpreteren dan "dat hele einde van de wereld ding is een zorg voor later". Van Huffelen denkt blijkbaar dat ze de risico's en de snelheid van de ontwikkelingen beter in kan schatten dan de duizenden AI researchers / CEOs die de alarmbellen laten luiden. Of ze ziet de risico's als acceptabel. Ik vind een gemiddelde van 14% kans op uitsterven niet echt een acceptabel risico, maar daar zullen de meningen over verschillen. Ik denk dat we hier het typische struisvogel effect zien. Het probleem is te groot en te eng om serieus te nemen, dus we negeren het.

Dan over urgentie: de ontwikkelingen gaan kei- en keihard. In 2022 was er een survey die AI-onderzoekers vroeg wanneer er voor het eerst een prijswinnend boek door AI zou worden geschreven. Het gemiddelde antwoord: 2039. Zojuist verscheen de nieuwe inschatting: 2029. Maar hier komt de echte shock: eind december 2023 won een Chinese professor een literaire award met een AI geschreven boek. Zelfs de experts onderschatten de snelheid. Van Huffelen onderschat het nog meer. We kunnen het ons niet permitteren om het tempo te onderschatten, want dan gaan we misschien allemaal dood.

Dennis, alsjeblieft, neem een uurtje de tijd om dit onderwerp met me te bespreken. Als je denkt dat het probleem onzin is, wil ik je vragen om je terug te halen hoe de meeste mensen dachten over klimaatverandering 15 jaar geleden. We zagen hetzelfde patroon: een groep wetenschappers slaat alarm, het publiek wil het allemaal maar niet geloven (is wel zo comfortabel) en de kranten staan vol met twijfel-zaaiende verhalen waardoor de politiek vrij spel heeft om absoluut niets te doen. We kunnen dit niet opnieuw laten gebeuren.

Mag ik je binnenkort trakteren op een lunch / koffie? Of liever even digitaal?
spiral steeple
#

Translation:

 Dear Dennis,

 How are you doing?  I would still really like to talk to you about suing the state for not doing enough to address the risks of AI.

 Last week, the government shared a vision on AI.  It's nice that the government is doing something, but in terms of content it means very little.  The biggest risks of AI (that literally anyone can die) are not explicitly mentioned, let alone addressed.  Van Huffelen recognizes "concerns mentioned in the call from tech celebrities", but notes: "however, the call focuses largely on future developments" and that there are already risks that "will not be resolved with a pause".

 Of course, that's not the whole point of a pause: it's not about the risks of current models, but about the risks of models that will come in the future.  That's why 33,000+ people are signing the pause letter, that's why countless AI professors say the existential risk must be addressed.  That's why the three most published AI scientists have all changed their careers to warn people.

 I cannot interpret the view of our cabinet other than "that whole end of the world thing is a concern for later".  Van Huffelen apparently thinks that she can estimate the risks and the speed of developments better than the thousands of AI researchers / CEOs who are sounding the alarm bells.  Or she sees the risks as acceptable.  I don't think an average of 14% chance of extinction is really an acceptable risk, but opinions will differ on that.  I think we see the typical ostrich effect here.  The problem is too big and scary to take seriously, so we ignore it.
# 
 Then about urgency: developments are going very fast.  In 2022, there was a survey that asked AI researchers when the first award-winning book would be written by AI.  The average answer: 2039. The new estimate has just been published: 2029. But here comes the real shock: at the end of December 2023, a Chinese professor won a literary award with a book written by AI.  Even the experts underestimate the speed.  Van Huffelen underestimates it even more.  We cannot afford to underestimate the pace or we might all die.

 Dennis, please take an hour to discuss this topic with me.  If you think the problem is nonsense, let me ask you to recall what most people thought about climate change 15 years ago.  We saw the same pattern: a group of scientists raises the alarm, the public doesn't want to believe it all (which is comfortable) and the newspapers are full of doubt-mongering stories, giving politicians free rein to do absolutely nothing.  We can't let this happen again.

 Can I treat you to lunch/coffee soon?  Or would you rather go digital?
lethal ivy
#

So Alex Champandard and I are actually preparing to sue a national government, with inspiration from the Urgenda case.

Here the focus is on how national governments did not prevent data laundering (starting with copyright violations). We pitched our case to French law master students last week.

winged patio
#

Gabriel Weil is a law (assistant) professor who wrote a paper on using "Tort Law" for suing AI labs for catastrophic AI risks.

Paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4694006

Podcast: https://www.youtube.com/watch?v=hdQDZo7jGBY

I've reached out to him on twitter.

YouTube
AXRP
28 - Suing Labs for AI Risk with Gabriel Weil

How should the law govern AI? Those concerned about existential risks often push either for bans or for regulations meant to ensure that AI is developed safely - but another approach is possible. In this episode, Gabriel Weil talks about his proposal to modify tort law to enable people to sue AI companies for disasters that are "nearly catastrop...

▶ Play video
winged patio
winged patio
#

Just chatted with Tyler Whittmer (from LASST) and Laurens Prins on what can be done in the US!

winged patio
lethal ivy
#

His integrity about acting according to what needs to happen is inspiring. The part where he says that even though this is a problem that involves everybody, that no-one would take his concerns seriously if he does not do what it takes – resonates for me.

winged patio
#

@silk needle is a dutch lawyer who is very much interested in this!

lethal ivy
#

Yes, he's been talking about it! Would be cool if we can actually get this to happen

weary eagle
#

Going to link up the AI plans Law-a-thon. More expert lawyers etc. Envisage any useful interaction @winged patio ?

#1186777799890059425 message