Olivia Guest · Ολίβια Γκεστ's Avatar

Olivia Guest · Ολίβια Γκεστ

@olivia.science.bsky.social

https://olivia.science assistant professor of computational cognitive science · she/they · cypriot/kıbrıslı/κυπραία · σὺν Ἀθηνᾷ καὶ χεῖρα κίνει

17,259 Followers  |  2,114 Following  |  6,994 Posts  |  Joined: 02.05.2023  |  2.2435

Latest posts by olivia.science on Bluesky


you sound like a dipshit libertarian who just finished their first semester of freshman econ 101 and has everything figured out

02.01.2026 03:46 — 👍 229    🔁 10    💬 3    📌 2

"DATA CENTER WATER USE DOESN'T MATTER" is one of the silliest arguments I see floating around. unless you think re-writing 200 yr old water law (OG political third rail in the US) is a simple matter, the marginal consumption of new water is literally make or break for much of life in the western US

02.01.2026 03:41 — 👍 997    🔁 211    💬 18    📌 34

Geniuses

23.02.2026 21:40 — 👍 1    🔁 0    💬 0    📌 0

Anthropic was supposed to be the "good" one, right?

23.02.2026 13:36 — 👍 50    🔁 17    💬 3    📌 0

"What can we do?

Zooming out, there is much you can do and think about — even just refusing to accept [the artificial intelligence] industry frames and rejecting their products is non-negligible."

23.02.2026 21:31 — 👍 10    🔁 3    💬 0    📌 0

Keep preaching

23.02.2026 21:28 — 👍 1    🔁 0    💬 0    📌 0
39C3 - AI Agent, AI Spy
YouTube video by media.ccc.de 39C3 - AI Agent, AI Spy

The move toward the "agentic" branding seems to be designed to gull people into thinking there's some new architecture out there, but it's just multiple LLMs sutured together, ass-to-mouth, under the hood.

I keep reposting, but if you don't want to sleep at night watch this Whittaker presentation:

23.02.2026 16:07 — 👍 11    🔁 4    💬 2    📌 0

Agentic technology appears to involve building a machine to produce strings of text that resemble things a person might want to tell their computer, and giving that machine control of your computer. Why would you.

23.02.2026 15:56 — 👍 9    🔁 3    💬 1    📌 0

A "rookie mistake."

Let me repeat for the thousandth time: "agents" are a nonfunctional fantasy technology that is, and will continue to, fail. The underlying LLMs are simply not capable of powering what is being asked of "agents;" this was not a technological development, it was a fantasy.

23.02.2026 15:54 — 👍 44    🔁 18    💬 3    📌 0

world-historical levels of wholesomeness

23.02.2026 20:11 — 👍 722    🔁 84    💬 10    📌 1

Funny how Meta executives let Google read their e-mails.

23.02.2026 21:14 — 👍 9    🔁 1    💬 1    📌 0
Preview
We've been here before! Parallels between AI and tobacco, and other warnings.

Also if useful olivia.science/before

23.02.2026 20:54 — 👍 2    🔁 0    💬 0    📌 1

Self parody

> “Rookie mistake tbh,” Yue said in another post. “Turns out alignment researchers aren’t immune to misalignment. Got overconfident because this workflow had been working on my toy inbox for weeks. Real inboxes hit different.”

23.02.2026 20:53 — 👍 17    🔁 3    💬 3    📌 0

A single tear runs down my cheek.

23.02.2026 18:04 — 👍 39    🔁 7    💬 2    📌 1

Uh, just going to point out that there's been zero progress on AI building "complex software," and AI cannot, and will not, solve "advanced physics problems."

These claims are lies, plain and simple.

23.02.2026 16:28 — 👍 61    🔁 19    💬 1    📌 0

“Computer literacy. Internet literacy. Social media literacy. Mobile literacy. Virtual reality literacy…The pitch to train schoolchildren on the latest tech has stayed roughly the same since the introduction of personal computers in the late 1970s…”

And yes indeed, we do fall for it every time.

23.02.2026 13:01 — 👍 44    🔁 17    💬 0    📌 0

Sent from my palantir

31.01.2026 15:37 — 👍 38    🔁 1    💬 0    📌 0
Preview
Meta Director of AI Safety Allows AI Agent to Accidentally Delete Her Inbox Meta Superintelligence Labs’ director of alignment called it a “rookie mistake.”

NEW: Meta’s director of AI safety, supposedly the person at the company who is working to make sure that powerful AI tools don’t go rogue and act against human interests, had to scramble to stop an AI agent from deleting her inbox against her wishes...

23.02.2026 15:23 — 👍 1139    🔁 446    💬 46    📌 200

but also, it's not that good at summarizing

23.02.2026 19:48 — 👍 194    🔁 37    💬 9    📌 2
Video thumbnail

Don't Hug Me I'm Scared is coming to @dropout.tv

23.02.2026 18:47 — 👍 743    🔁 200    💬 15    📌 50
FWIW, here are my predictions about what will happen to twitter:

1) "Free speech" will make it an information cesspool like found in newspaper comment sections. The societal effects will be very bad, but through strategic blocking some people will manage to be ok.

FWIW, here are my predictions about what will happen to twitter: 1) "Free speech" will make it an information cesspool like found in newspaper comment sections. The societal effects will be very bad, but through strategic blocking some people will manage to be ok.

I've been off twitter for a while but kept it so nobody would take it. Going through it, I found this.

Me (and a lot of other people) predicted what Elon's takeover would do to our information ecosystem. I remember being told I was overreacting and fearmongering. hmmm 🤔

31.10.2024 21:41 — 👍 28    🔁 3    💬 2    📌 1

Regarding all the evidence showing X is in fact an efficient Nazi propaganda factory... well, some of us who study propaganda and disinformation did in fact predict this, only to be told we were overreacting.

The quoted thread contains screenshots of predictions I made back in October 2022.

23.02.2026 11:06 — 👍 31    🔁 10    💬 1    📌 0

Don't forget the users. They must be made accountable, too.

23.02.2026 19:30 — 👍 9    🔁 2    💬 0    📌 0
Post image

Every now and then, I remember that JD Vance smashed the "like" button after reading this tweet, then went back and unliked it five years later when he ran for the Senate.

23.02.2026 16:44 — 👍 3850    🔁 767    💬 30    📌 13

From: which.king@angmar.gov
To: xXSauronDaDeceiverXx@angband.edu
Date: 1554 Second Age
Subject: Re: Master Ring (Forged in secret)???

lmfao u totally shoulkd

31.01.2026 15:34 — 👍 927    🔁 235    💬 11    📌 3

DOCTOR: I told him he needed to get out to a show, that was how he would cure his depression.

CHOTINER: So you learned this technique in school?

DR: No, not— listen it was good advice. Pagliacci was in town.

C: Right. Is it standard to give advice before learning a patient’s name?

DR: Now look

02.02.2026 03:58 — 👍 8716    🔁 1796    💬 1    📌 35
A two-panel Little Bubbie Child comic. Top panel is the little bubbie child asking, "Dad, hwat was it like a livin' in the 90s?" Bottom panel is dad responding, "Sometimes you'd open a can a peanuts and they'd come snakes outta there."

A two-panel Little Bubbie Child comic. Top panel is the little bubbie child asking, "Dad, hwat was it like a livin' in the 90s?" Bottom panel is dad responding, "Sometimes you'd open a can a peanuts and they'd come snakes outta there."

10.02.2026 15:55 — 👍 8258    🔁 1594    💬 6    📌 34
William Blake’s illustration for “The Tyger” has what appears to be a dippy grin for some reason.

William Blake’s illustration for “The Tyger” has what appears to be a dippy grin for some reason.

The wonderful thing about Tyggers
Is Tyggers are burning bright
Their tops are made out of rubber
In the forest of the night

What bouncy, trouncy, flouncy, pouncy,
Immortal hand or eye
Did frame the most wonderful thing about Tyggers,
Their fearful symmetry!

04.02.2026 02:13 — 👍 1069    🔁 426    💬 18    📌 19
Video thumbnail

Wow Signal

17.02.2026 18:48 — 👍 8586    🔁 3626    💬 85    📌 32
Post image

*smiles insanely* it’s perfect

22.06.2023 14:21 — 👍 1418    🔁 285    💬 9    📌 10

@olivia.science is following 20 prominent accounts