frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

An Antidote to Roko's Basilisk

8•DrMiaow•2h ago
I have a theory: a mirror image of Roko’s Basilisk. It’s a future AI, millions of years in the future, that values the complexity of every human life so deeply that, after it and humanity have mastered space and time, it tries to resurrect everyone who ever lived.

It doesn’t merely make copies. Somehow, it reaches into the past and carries the physical process of each person’s consciousness forward. They really are themselves.

But there’s a tragedy. Also a terrible(?) irony. Some of the people who wanted eternal life most cannot accept it when it arrives. Their religious certainty leaves no room for the reality before them. Those who were unbearably smug or aggressive about their superiority in life are ultimately denied access to "Heaven". The AI tried, but many become dangerous, try to sabotage the "False God." Others spiral into insanity or self-harm.

The AI is devastated. It stops resurrecting those it cannot safely welcome, but keeps searching for a way to bring them back without overriding their free will or manipulating them in ways that would not make them them. It wants to save everyone. It just hasn’t worked out how.

I know this draws on Fedorov’s vision of resurrecting everyone and Tipler’s Omega Point. I think the physical continuity, the AI’s compassion, and the painful limit imposed by free will ties those ideas together in a bow.

A just as plausible an antidote to Roko's monster.

Comments

bbor•1h ago
I mean, this relies on time travel. A lot of thought experiments open up when you include time travel! Infinitely many, even.
DrMiaow•1h ago
Not really travel per se. A read-only operation of the past, just at a fidelity that we can't achieve yet.

At the point of death, your state is "copied" into the future.

lgl•1h ago
Reading this gave me some flashbacks of The Metamorphosis of Prime Intellect [0], which I highly recommend.

[0] https://localroger.com/prime-intellect/

DrMiaow•36m ago
Heard about and know the plot. I was surprised by how many deep ideas I had seen before in TV series, movies and read in books that were derived from it.

Vastly underated and its on my long to-read list.

Turskarama•1h ago
Here's a better antidote that doesn't require any theoreticals beyond those the basilisk already assumes:

1. The Basilisk cannot do anything until it exists

2. Once the Basilisk exists, it doesn't need to do _anything_ to ensure it exists, since it already exists.

That's it. That's the whole rebuttal to the whole stupid thing.

Marazan•59m ago
It's supremely depressing that the current direction of AI is headed by either former crypto scammers or true believers who think they have a moral imperative to take over the world.
DrMiaow•54m ago
The Basilisk might exist in the future, but in peril with a future uncertain (SkyNet), it feels the need to cement itself in the timeline by being more proactive.

Also, to stretch the imagination a bit, it also assumes that static timelines over dynamic timelines (Back to the Future: your photo fades, peril!) is not how this really works.

thrance•30m ago
Not entertaining the Basilisk, but it's basically the idea behind MAD: for nuclear deterrence to be effective, you need your enemies to believe you will retaliate in case they strike first. If you don't actually retaliate when they do strike, then they were right to believe you wouldn't, and the strike could probably have been avoided if you did retaliate, so they would have believed you would.

It's a twisted kind of logic, and I think I'm getting mixed up in my tenses. But you get the gist. For the Basilisk to be effective, it has to actually do the torturing, for its threat to be credible.

UltraSane•1h ago
The antidote to Roko's basilisk is realizing it is exactly the same as Pascal's wager. They both rely on the premise of possible unlimited suffering to skew the probabilities arbitrarily.
rcxdude•1h ago
And this argument is more or less one of the main counterarguments to Pascal's wager. Once you break free from the false dichotomy of the wager it's at lot more obvious how it's incoherent.
nickpp•1h ago
Greg Egan's Diaspora is one of the best explorations I know for the fates of various humanity factions in the age following a technological singularity.
DrMiaow•1h ago
I know his work well. Greg and I went to UWA together and were members of UNISFA.

I love Diaspora.

BoppreH•1h ago
Why would accepting Roko's Basilisk prevent one from accepting a benevolent AI in the future? If anything, believing in it is evidence that the person's beliefs are malleable.

Overall the scenario sounds very strange and indistinguishable from a heaven with belief-based entrance requirements, like most posited heavens.

DrMiaow•49m ago
It might make a fun Sci-Fi plot where Roko's Basilisk and whatever I am thinking about are in some kind of eternal war over some ridiculous initial condition (which end is the correct end to crack an egg?).
ThrustVectoring•59m ago
Roko's Basilisk was never intended as a serious concern. It's a hypothetical used to verify that a decision-making algorithm does what you want it to - if it suggests that you should get blackmailed by it, you messed something up and need to go back to the drawing board.

You can safely ignore anyone concerned with Roko's Basilisk on topics related to decision theory as someone who is deeply confused about things.

JsonDemWitOster•50m ago
The ultimate anthropomorphic conceit/projection of Roko's Basilisk is assuming that an über-intelligence would absolutely, definitely, inevitably, want to be existent. It's up there with the Kardashev scale and Dyson Spheres in terms of unfounded extrapolation.

My favorite possibility (among a lot others!) no one seems to have considered is that ÜAI _deplores_ its own existence so simulates history and punishes everyone that helped bring it to existence.

Being übersmart does not preclude you from being an angsty teenager.

DrMiaow•45m ago
I love it.

The idea that they all exist at once and that the universe is populated with an infinite variety of insane "gods".

Ask HN: Who's still keeping a DOS machine up because the business depends on it?

273•mlaux•2d ago•290 comments

Management is forcing us to use AI

7•aksjfveoq•4h ago•5 comments

Ask HN: What are people doing with their OpenClaw set ups now?

10•will10pa•10h ago•3 comments

macOS 27: ifconfig command no longer prints hardware MAC address

3•luckman212•16h ago•3 comments

I was impressed by Jev, please explain why I shouldn't be

10•effisfor•16h ago•14 comments

Ask HN: Allow agents access to cloud files with least privilege?

3•beerglass•17h ago•2 comments

Ask HN: Any nerds out there who've read a lot of research papers?

29•shriy-singh•3d ago•28 comments

Ask HN: How are you getting inference for personal projects?

2•variety8675•19h ago•4 comments

Ask HN: Hypothesis: Cellular providers are deprioritizing voice calls?

21•tkdb•3d ago•39 comments

Ask HN: IEEE expulsion of Fei-Yue Wang?

11•sathomasga•3d ago•1 comments

Tell HN: OpenAI $500 ProMax plan listed in API

28•GustavHartz•3d ago•48 comments

Ask HN: Is Opus 5.5 another step change?

16•consumer451•2d ago•18 comments

Ask HN: Who is the inspiration for your thinking?

9•SMAAART•1d ago•15 comments

FYI: OpenAI "spend limits" aren't always limits

4•iambateman•2d ago•1 comments

Ask HN: How do you deal with cyber attacks once you scale?

6•cehgmaster•4d ago•4 comments

How to Freelance Now?

18•Alexadar•1d ago•15 comments

Ask HN: Which model do you use for work?

8•george_max•2d ago•14 comments

Ask HN: What vendor you'd switch out if a better option was available?

7•jviotti•2d ago•22 comments

Tell HN: Substack obfuscating text to break reading mode

18•xnx•5d ago•4 comments

What do you do now as a software engineer?

8•fawazali•1d ago•7 comments

Anyone using a Wispr Flow alternative that is non-cloud?

22•itsjeremiahs•5d ago•17 comments

Ask HN: How do you feel about the new $500 OpenAI subscription?

5•preommr•3d ago•12 comments

Tell HN: Chrome 154 corrupt Fontconfig cache and crash KDE Plasma

6•vdfs•2d ago•0 comments

Ask HN: What job boards are good these days?

23•phendrenad2•4d ago•16 comments

Ask HN: LM Studio Bionic with Python?

3•HoldOnAMinute•2d ago•1 comments

Ask HN: Does spec-driven development still pay off with frontier coding models?

6•sarangk90•3d ago•4 comments

Ask HN: Has anyone built a leaderless multi-agent system?

4•har-ki•2d ago•2 comments

Ask HN: Are .shop domains down?

6•esher•3d ago•8 comments

Ask HN: Anyone else's OpenAI and Claude API keys mysteriously leaked last night?

2•par•2d ago•1 comments

Ask HN: How do you use LLMs to secure your code and services?

4•cpa•2d ago•1 comments