The Five-Floor Goodbye, by Ryan Ike from the album Gunpoint - The Soundtrack

I'm sorry but

Windows Start Menu with "add" typed into the search bar. Results include "Add a printer or scanner", "Madden NFL 25", "Device Manager" and an option to search the Web for "add remove programs" The actual system settings page "Add or remove programs" is not listed

MICROSOFT?

Microsoft Windows Start Menu. "remove" has been typed in the search box. Search results several directories named "remove", "Delete temporary files", "Check firewall status", "Delete browsing history", "Delete cookies or temporary files", and of course an option to search the web for "remove programs" The system settings page "Add or remove programs" is not listed as a result

HELLO? ADD OR REMOVE PROGRAMS?

no children no background no ID no style

"Interactions with this person do not spark joy"

🐦 Thank you! Tell your most influential friends!

Go away bot

Two perfect AIs cannot understand (model) *each other* completely because that would imply that they understand *themselves* completely which would involve an infinite regression, which is impossible

This is nothing to do with the Halting Problem?

What does the Halting Problem have to do with this

But it's circular, it involves an infinite amount of analysis/information

This thread has been pretty unfocused, I was mostly thinking out loud, but I think one solid takeaway would be that intelligence, even infinite intelligence, is not the be-all and end-all in most real-world scenarios

A chicken pecking a red button can switch off a supercomputer

Yes!

In the Culture books there's an upper limit to an AI's intelligence, but perfect AIs always sublime (ascend to a higher plane of reality)

Which is an incredibly cool way to ensure that the remaining hyperintelligent AIs in that universe are heterogeneous and have personality and disagreements

This sounds reasonable at first... but what if "everything" includes another AI elsewhere in the same universe which is engineered in exactly the same way as this one? They can't *both* take one another perfectly into account

Or take *themselves* into account

You are correct

Is the entire universe rock-paper-scissors?

Especially if "build a better AI" is one of the moves that side B could make, which side A has anticipated and taken into account

So success is more about physical resources, and speed of movement

I'm imagining a war where the belligerents both have access to AIs capable of gaming out all of their opponents' possible moves and responses, to infinity, with the result that

1. both sides conclude that side A will win
2. no amount of additional CPU for side B can EVER change that reckoning

It's also pretty easy to concoct physical real-world scenarios where it doesn't really matter how intelligent you are, you'll still lose, you've lost before the "game" "begins"

If "intelligence" is simply "an ability to simulate your opponent's behaviour in compressed time" then there's no upper limit to how far you can go with that... but it still doesn't necessarily result in a win (chess being a prime example)

Suppose you build the most intelligent AI in history. Which of these statements is true:

1. It's always possible to build a more intelligent AI which can outsmart the first one
2. It may be impossible to get a better result than stalemate

"Intelligence" is something I'm inclined to put into scare quotes because it has a lot of different axes, some of which do have well-defined limits and some of which I feel are a lot more like rock-paper-scissors

Take IQ, however you want to define "IQ" - is there a ceiling there? "150 IQ" means something. Do we reach a point where "1000 IQ" stops meaning anything? No matter how much CPU you throw at it no matter how huge and energy-hungry the brain, there's a maximum possible amount of "smart"

So there's an upper limit to how good a chess AI can be at chess. There is a (highly theoretical) perfect strategy and an AI could play that strategy. If both sides play that strategy the game could well end in a draw

Is there an upper limit to "intelligence" generally

Time-based one-time password - Wikipedia

Time-based one-time password

en.wikipedia.org/wiki/Time-ba...

Configuring two-factor authentication | npm Docs Documentation for the npm registry, website, and command-line interface

Here is the relevant documentation stating that this should just work

npm 2FA/OTP Issue: Unable to Retrieve One-Time Password · community · Discussion #181802 Select Topic Area Question Body Hello, I have enabled Two-Factor Authentication (2FA) on my npm account. However, when I try to run npm publish, I am prompted to "Enter one-time password", but I am...

4. The best known solution to this problem is to specifically bypass 2FA?

npm notice Publishing to https://registry.npmjs.org/ with tag latest and default access npm error code EOTP npm error This operation requires a one-time password from your authenticator. npm error You can provide a one-time password by passing --otp=<code> to the command you ran.

3. When attempting to publish on the CLI using a passkey they still *prompt* for a TOTP which is now impossible to obtain?

Unable to Enable 2FA with Authenticator App - Option is Missing · community · Discussion #182325 Select Topic Area Bug Body Hi everyone, I'm trying to set up two-factor authentication (2FA) on my npm account, but I'm unable to find the option to use an authenticator app (like Google Authentica...

2. TOTP support has been removed even though passkeys do not work?