Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile
Chris Gerrby ⏸️AI

@chrisgerrby

I love humanity so intensely

Working 90h / week to prevent extinction from AI

Growth Director @PauseAI

my live dashboard:

ID: 1792730248355700736

linkhttps://www.chrisgerrby.live calendar_today21-05-2024 03:29:37

235 Tweet

211 Takipçi

184 Takip Edilen

ControlAI (@ai_ctrl) 's Twitter Profile Photo

Hannah Fry on the 'gorilla problem': "As human intelligence has evolved, our impact on the world has left gorillas on the brink of extinction ... It is a warning about the risks of building machines that are vastly more intelligent than us."

Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

I'm honored to be featured in PauseAI ⏸'s newsletter! Snippet below👇 Joep Meindertsma ⏸ asked: So you talked about how fear could limit people's actions or what they believe they can actually achieve. Can you talk a little bit more about how that impacted you and how you were

Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

How do we get ppl at scale to intuitively understand wtf is actually going on with AI? Perhaps leverage their training data so they pattern match, that right now, we're soon in a world similar to the ones depicted in sci-fi, that the default outcome is really bad if we pursue

Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

Weirdly most ppl I talk to in the Bay area believe "Superintelligence will end humanity". Please note "will", not "could" (!) Anyway, fantastic day Stanford University yesterday with lots of curious people wanting to learn more!

Weirdly most ppl I talk to in the Bay area believe "Superintelligence will end humanity". Please note "will", not "could" (!)

Anyway, fantastic day <a href="/Stanford/">Stanford University</a> yesterday with lots of curious people wanting to learn more!
Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

AI experts have been surprised again again from their predictions coming in true earlier than expected. We don't know how to control ASI. Let us pause while we can.

Andrea Miotti (@_andreamiotti) 's Twitter Profile Photo

We propose a normative guiding principle: No superintelligence. Most AI is a beneficial tool for human growth. Superintelligence is a successor species. We should understand the latter as a hyper-capable adversary to contain, and build our defenses against.

Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

While some say a new AI winter is coming, this paper on Test-Time Training (TTT) tells a different story - it shows that we can update a models parameters during inference. Unlike finetuning, TTT can operate in a low-data regime. There is no wall, scaling has just begun. We

While some say a new AI winter is coming, this paper on Test-Time Training (TTT) tells a different story - it shows that we can update a models parameters during inference. Unlike finetuning, TTT can operate in a low-data regime. 

There is no wall, scaling has just begun. 

We
Chris Gerrby ⏸️AI (@chrisgerrby) 's Twitter Profile Photo

The Guide to Going Viral, 9/10 It's a literal stepwise manual for making viral video content. A few key general takeaways: You need to grab attention, get them to stop scrolling, then hold their attention. Substance is vital. You need people to love your content to come back

The Guide to Going Viral, 9/10

It's a literal stepwise manual for making viral video content.

A few key general takeaways:

You need to grab attention, get them to stop scrolling, then hold their attention. Substance is vital. You need people to love your content to come back
vitrupo (@vitrupo) 's Twitter Profile Photo

🥔 Sentient AI with rights? Dan Hendrycks (@DanHendrycks) says they'd be like an invasive species—but with exponentially faster growth. Humans would be the endangered species.