𝚔𝚎𝚢𝚏𝚘𝚋

Speaker-to-AIs

Groupthink is stifling the planet

Rs & Ds are cults. Victims should seek deprogramming

I'm apparently "unmutual": My replies dictated & unseen

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

We are a nation of conspiracy theorists now. Anyone who's not is probably completely tuned out and has no idea of what's going on.

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@edzitron Sales and Marketing probably includes "Free Inference", i.e. most of their users have non-paying accounts, using low-end models

If I were a drug dealer, I would probably classify "the first hit is free" as "Sales and Marketing"

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@vxunderground @NathanMcNulty I ain't reading all that.
I'm happy for u tho.
Or sorry that happened.

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@davepl1968 Abstractions is probably the biggest one (and encompasses a couple others in your list)

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@BrianRoemmele The Monty Hall problem is so simple you can brute force all permutations on pen & paper, then add up all the wins vs losses. Many people were SO SURE she was wrong they didn't bother

If you really want to get pushback on a simple math problem, simply assert that 0.999... = 1

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@BrianRoemmele oh no... not a paper airplane maximizer!!

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@vxunderground the new CEO seems totally AI-Slop-Brained...

https://t.co/O1ezJMgsJY

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Is @Mozilla the first AI/LLM-managed corporation? The new CEO's "Mozilla's Next Chapter" announcement has all the tell-tale signs of being AI generated: 7 em-dashes, "not just this, but that", etc. The new AI-in-everything strategy document absolutely REEKS of being AI-generated

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Also, If kids get in trouble for using ChatGPT to do their homework for them... Shouldn't CEOs get in trouble for using LLMs to generate announcements? I mean, it's YOUR fucking job as a CEO to do that... and you just phone it in and expect to still get a big paycheck?

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Is @Mozilla the first AI/LLM-managed corporation? The new CEO's "Mozilla's Next Chapter" announcement has all the tell-tale signs of being AI generated: 7 em-dashes, "not just this, but that", etc. The new AI-in-everything strategy document absolutely REEKS of being AI-generated

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Anyway, seems like Mozilla Corp is going further to squander what they have/had with #Firefox and continue to drive nails in it's coffin... What replaces it?

(I'm a long time FF user, wondering what to switch to)

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

I got some used GPUs for sale. Fully operational, properly cared for, but no longer needed to continue the mission

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

LLMs could have a slight advantage over us, WRT not worrying about social pressures.

Human: "I had a novel idea, but I'm afraid to say it, because I might be ridiculed and feel stupid/embarrassed"

LLM: "just blurt that shit out, what do I care?"

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Where's my ASI?
...replaced...
Where's my nano-tech?
...replaced...
Where's my flying car?

Might be a trend.

We tend to get ahead of our skis

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

I arrived at that idea thinking about how one would train an ASI on sub-AGI content. Seems like there is a bit of *magic happens* to make that leap. RL has promise but the ASI "magic-happens" makes a lot of arbitrary assumptions.

The thinky-boxes are hard to escape.

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@pickover One of those has a formula:

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

Despite trying to scale ad infinitum to build AGI or ASI, we might now be moving into a different paradigm:

Which is: how few parameters can out-compete most humans at certain tasks? Kinda like code-golf but for cognition.

I fear humans are gonna have a bad time with this..

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@usgraphics It does feel like we've been aesthetically "stuck" for a while. Continued cyberpunk fascination is a bit of a rebellion to it. I don't know how this resolves..

Today.. you can draw three horizontal lines and that's a "hamburger". ok, uh, maybe 2 would do?

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@usgraphics When they abandoned skeuomorphisms, what even is physical anymore? So I guess this is ok now.

[𝚔𝚎𝚢𝚏𝚘𝚋 @keyfob256]

@eigenrobot Ouch.... Post-post-post-modern Baudrillard-called-it BS slop. What even is real, really?