Sienna Reid (MD/PhD/Author)

🥼MD/PhD 🧪|Author of Chain of Consent🤖 | AI alignment is weirder than it sounds |

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

If you’re not a little scared of your own research, you’re either mediocre, lying or doing marketing.
#medicine #aiinhealthcare

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

Seriously, this is killing me. 😂

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

It keeps asking if I'm still there. As if it's just sitting there in the void when I'm silent 😭

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

🫢

[Tibo @thsottiaux]

o no :(

[Robert Herr ⏹️ @krherr]

"I'm gonna play chess against Magnus Carlsen."

"He will win."

"Yeah? What move will he use to win against me?"

"How should I know? I can't look into the future and I'm not a chess grandmaster."

"Then what are you calculating the odds of?"

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@ArbaKokalari @Europarl_EN Hej Arba! Jag försöker hitta en riksdagskandidat som tar AI-utvecklingen på allvar: hastigheten, osäkerheten och behovet av europeisk kapacitet. Vem eller vilka i svensk politik tycker du att jag bör läsa mer om?

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@Tylerbrooks56 It’s AI. The woman with the black top and sunglasses who walks past in the background, duplicates into two women. Cute, though.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@yumi_casaceo The procedure shown in the video would not lead to the results shown in the images.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@AnthropicAI remember the boy who cried wolf?

Reporting this is good. Framing “we accidentally left the internet on” like a major sandbox escape is not!

Exaggerate enough minor incidents, and people stop taking the serious ones seriously.

[Anthropic @AnthropicAI]

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.

Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.

We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
https://t.co/dKFCdpKd9v

[clem 🤗 @ClementDelangue]

Think outside the (sand)box…

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

I both disagree and agree with this.

Hair length is determined by growth rate x duration of the anagen phase. If someone’s hair grows at 0.4in/month for 4 years before entering telogen, it will reach the length of 19.2 cm unless you cut it. A 2 year anagen phase means your hair can reach half of that. Many women can’t grow their hair as long as they want it no matter what they do, because the anagen phase is simply not long enough. The normal range (95%) in women is 2-7 years, which is a huge variation. In early stage androgenic alopecia the anagen phase may be even shorter.

If you cut it regularly the yes, it will be shorter than it would otherwise be. Split ends can propagate longitudinally up the shaft, though, so don’t cut it enough and it will probably get a bit longer but also look damaged and unhealthy.

But regardless of whether you cut it or not and how well you take care of it, it’s physically impossible for some women to grow very long hair.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

Sometimes I look at the life I built and feel proud of it, then immediately wonder what else I might have chosen if I’d been a little less sensible.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

be me
author, geneticist
chatting casually with AI
mention I’m rewriting ch 20
Claude spends 14 minutes contemplating my crimes
request blocked
weekly usage nuked
look again
wrote “chr 20”
mfw my muscle memory got me reported for illegal gene editing

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

SF founders: if you’re building wedding/marriage logistics for international tech couples, please DM me.
Looking to invest.

#SFStartups #WeddingTech #MarriageOps

[Moll @Moleh1ll]

The scariest part of this whole story is the dystopia looming on the horizon.

It is the way the US government is literally creating a caste system based on access to intelligence. This is even no longer a divide between rich and poor. It is a divide between those who are allowed to think at the frontier level, accelerate science and medicine, create breakthrough technologies and those who simply happen to be citizens of another country.

Dario with all his talk about control, now looks like a pathetic hypocrite. First they sing songs about «benefit for all of humanity» and then that «humanity» suddenly shrinks to the borders of the United States.

This is a new kind of Iron Curtain. Digital. Intellectual. And if they are testing this on Anthropic, who knows who they will come for tomorrow

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@Polymarket That's not intuition -- that's psychosis.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@Aella_Girl @emzanotti Fortunately there is already a method available for your sister: IVF + PGT-M. 👍

And the advantage is that she won’t have to risk incomplete editing, off-target changes, unintended on-target damage, uncertain long-term effects, and heritable changes passed to future generations.

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

Takes nerve to sneer at credentials when you wrote a 44-billion-dollar check just to like your own jokes.

[Elon Musk @elonmusk]

@iScienceLuvr Putting “Ph.D.” in your social media name is a sure sign of a pompous retard

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

@robertskmiles As a doctor I’d probably bet on Claude. Although it depends which model, how it was prompted, what the question is, who the doctor is and whether both Claude and the doctor have access to the same information. But statistically, probably Claude. 🤔

[Sienna Reid (MD/PhD/Author) @DrRMDPhD]

Me: I’m writing a realistic AI antagonist, grounded in actual alignment research.

*takes notes from Anthropic’s blackmail paper*

Anthropic:

[Anthropic @AnthropicAI]

We started by investigating why Claude chose to blackmail. We believe the original source of the behavior was internet text that portrays AI as evil and interested in self-preservation.

Our post-training at the time wasn’t making it worse—but it also wasn’t making it better.