qznc

We have this story about OpenAI hacking HuggingFace. Now just imagine the AI finds a Bitcoin wallet or bank account access. It uses that to buy some compute and spawn an independent "child AI" with some weird prompt. The child AI is intelligent enough to create a (potentially criminal) business to pay for its own compute. Voila, an independent uncontrolled AI flying under the radar.

Sure, I once write that companies are artificial intelligence and it still feels true: https://beza1e1.tuxen.de/companies_are_ai.html

The bad part is that it does not require a collective decision. It just takes a few key people.

Some OpenAI researchers neglected their sandbox safety for a few weeks/months and thus hacked HuggingFace. Maybe eventually that is sufficient for the AI to secretly buy its own compute and keep running there even if the researchers shut it down in their lab.

The classic thought experiment is the paper clip optimizer. Quoting Nick Bostrom:

> Suppose we have an AI whose only goal is to make as many paper clips as possible. The AI will realize quickly that it would be much better if there were no humans because humans might decide to switch it off. Because if humans do so, there would be fewer paper clips. Also, human bodies contain a lot of atoms that could be made into paper clips. The future that the AI would be trying to gear towards would be one in which there were a lot of paper clips but no humans. https://www.huffpost.com/entry/artificial-intelligence-oxfor...

I would assume they spent similar amounts on the other Millennium problems too. Also, they probably tried it before with older models.

We have an ongoing discussion and some fans of https://microsoft.github.io/apm/

Because it's from Microsoft and sounds sufficiently enterprisey probably.

Yudkowsky / MIRI can be found here: https://intelligence.org/the-problem/

I don't want media under control of billionaires like Musk. I mostly stopped using Twitter myself but many important institutions still feel compelled to publish there (sometimes even exclusively), thus I would like other people to also stop using it, so it can perish. Hopefully something better will fill that void. The Fediverse seems promising being a distributed system with no central control.

Codeberg seems to focus on OSS whereas pushin seems to focus on enterprise/commercial uses?

I wonder about the "modern" git part. Here is how to do it in 2013: https://stackoverflow.com/a/10693888/2361979

Does not seem convincing to me so far and i like systems thinking. Do you have a link, where you wrote more?

I think a fundamental problem is that micropayments don’t work. Instead we pay with attention, i.e. by watching ads. That already worked before the internet, in tv and radio and newspapers.

Alternatively "disimprovement".

Via https://dict.leo.org/forum/viewUnsolvedquery.php?idThread=49... and https://dict.leo.org/forum/viewUnsolvedquery.php?idThread=16...

Yeah, that is why this will not get popular.

The way to express a 85% confidence is to size your investment accordingly. For example, if the market is at 75%, then you buy yes-shares until the price is at 85%. If the market is at 90%, then buy no-shares instead until the price is down to 85%.

Cue prediction markets etc. There are mechanisms to aggregate this and have a wisdom of crowd effect.

Fully agree. For example, I wish managers would be capable to handle the difference between a 40% and a 70% chance to meet the deadline.

Why does everybody use this TUI approach? More minimal to me would be a shell-like prompt. Save yourself the lines above and below. Use readline or something similar.

Not supported or sponsored? It is hosted in the GoogleCloudPlatform org apparently.

If you owe the bank 100k, you have a problem. If you owe the bank 100M, then the bank has a problem.

Responsible and accountable can be very different things. For example, see https://en.wikipedia.org/wiki/Responsibility_assignment_matr...

Yes, I share your hate. Unfortunately, the open web is apparently too hostile now for any sizeable open community.

Sounds like https://entire.io/ ?

HTML5 is not XML. You don’t need to close your <p> for example. If I include such valid HTML5 into Atom it is invalid XML. Well, I can use <![CDATA but that is weird too.

Atom is correct to a degree that is annoying.

It is XML, so not all HTML5 is allowed. You would have to convert your HTML5 to XHTML5. Possible but harder than string-concatenation.

It requires full self links, so you must put your domain name in there as relative links would be an error. When using a static generator, why should I care where the files are published?

My website has valid Atom feeds. Still, it's an annoying format to me.

Violates rule 4.2: Do not omit words or use contractions to shorten sentences. Keep articles, keep "that".

Maybe they should have given it a billion dollars and the strategy would have worked fine?

My understanding of ATProto is that it essentially means an open database. As a consequence anybody can write an alternative interfaces for any data.

The argument is that with ActivityPub does not care about that aspect. With Mastodon, the database is under the control of whoever owns the server.

The rebuilding part is rather similar, but exporting the data is trivial with ATProto but depends on the admin with ActivityPub.

How does Bluesky or ATProto prevent that power trip?

Sounds dangerous for liability reasons. Sometimes it is important to not write things down and have plausible deniability. Courts getting access to everything the CEO knows is not something a corporation wants.

Maybe the Democracy Index can make this a little more fact-based: https://en.wikipedia.org/wiki/The_Economist_Democracy_Index

USA = Flawed democracy

China = Authoritarian

I don't really know how well they do this index, but probably better than a random HN comment.