Author | Lightnews

G_Dino @gprdino.bsky.social · Jun 21

"Where the AI really struggles right now is knowing where it's made a wrong turn."

The discussion on being superficially flawless and AI generations being "odorless" in terms of #codesmell or "mathematical smell" is right on the money.

#TerenceTao #GenerativeAI

The limits of AI in mathematics: What makes human special | Terence Tao and Lex Fridman

YouTube video by Lex Clips

www.youtube.com

G_Dino @gprdino.bsky.social · Mar 26

While not yet peer reviewed (& heavily reliant on self-reports); doesn't seem surprising that "affective" power-users on voice mode would be more emotionally invested & attached. The interface is conversation, after all. Hmm.

#GenAI #OpenAI #AI #ChatGPT

Early methods for studying affective use and emotional well-being on ChatGPT

An OpenAI and MIT Media Lab Research collaboration.

openai.com

G_Dino @gprdino.bsky.social · Mar 13

Interesting precedent.

Revealed: How the UK tech secretary uses ChatGPT for policy advice

New Scientist has used freedom of information laws to obtain the ChatGPT records of Peter Kyle, the UK's technology secretary, in what is believed to be a world-first use of such legislation

www.newscientist.com

G_Dino @gprdino.bsky.social · Mar 5

Hmm. I sense a coming market for genetically altered pets. As though selective breeding wasn't enough for us to fuck animals up.

To be fair they're unbelievably cute.

Source:
www.iflscience.com/colossal-cre...

G_Dino @gprdino.bsky.social · Mar 4

This is confirmed; and also why it's getting review-bombed on Steam. lol

It kinda feels like the Old School R-Type and Final Fight II games on SNES. You gotta just be slightly masochistic and driven to a fault. lmao

#DeltaForce #BlackHawkDown #Gaming

Delta Force Campaign Is So Hard, Devs Cannot Even Complete It Solo

The Delta Force Black Hawk Down camopaign is so hard, no one at Team Jade managed to complete it solo.

mp1st.com

1

G_Dino @gprdino.bsky.social · Feb 17

Science illusion: science & scientific language "as a brand, not necessarily in substance."

Goodness illusion: "the illusion of goodness doesn’t represent what the science says."

Opinion illusion: "opinions commodified and marketized and weaponized."

bigthink.com/thinking/sci...

Reposted by G_Dino

The New York Times @nytimes.com · Feb 11

More than 150 bombs from World War II have been found under a children’s playground in northern England, with concerns that more may remain, officials said.

More Than 150 World War II Bombs Found Under Playground in England

The devices found during construction work were practice bombs, which can be harmful. Officials said there could be more.

www.nytimes.com

19 49 280

G_Dino @gprdino.bsky.social · Feb 9

"The period between the early 1940s and the mid-1960s is called “the Golden Age of Antibiotics” ... (when) almost two-thirds of all antibiotic drug classes were developed... Most are still used today."

#Antibiotics

Source:
ourworldindata.org/golden-age-a...

G_Dino @gprdino.bsky.social · Feb 4

#Upskill to be able to surface genuine takeaways from the noise and synthesize relevant and coherent insights from contemporary silos.

G_Dino @gprdino.bsky.social · Jan 30

Seems pretty rad, really. More of these directions would be nice.

I tried Google's new Daily Listen feature and I kinda love it

Google has a brand-new news report powered by your Discover feed, so I decided to try it out. Here's what I like about Daily Listen.

www.androidauthority.com

G_Dino @gprdino.bsky.social · Jan 28

Hmmmmmm

G_Dino @gprdino.bsky.social · Jan 26

Dropped by Booksale and found a couple of gems lol

G_Dino @gprdino.bsky.social · Jan 24

“It remains an open question whether we need foundation models … that can engage in human-like conversations and write Shakespearean sonnets if what we want is reliable information useful for educators, healthcare professionals, electricians, plumbers and other craft workers.”

A new look at the economics of AI | MIT Sloan

MIT Institute Professor Daron Acemoglu predicts that artificial intelligence will have a “nontrivial, but modest” effect on GDP in the next decade.

mitsloan.mit.edu

G_Dino @gprdino.bsky.social · Jan 22

Love it. Would be awesome to see actual pine cones and these do the same thing at the same time of day and weather conditions.

Pine cone-inspired building shades open and close autonomously without electricity

Researchers have created a novel, energy-independent facade system that passively responds to weather changes.

www.yahoo.com

G_Dino @gprdino.bsky.social · Jan 22

This is gonna be great. I mean ChatGPT already speaks better than him so it's actually an improvement.

Trump Admin Accused of Using AI to Draft Executive Orders

Artifacts in Trump's executive orders have led to speculation that those who penned them might have turned to AI for help.

futurism.com

G_Dino @gprdino.bsky.social · Jan 20

Well look who's back. Every Frame a Painting. lol

#FilmMaking

Where Do You Put the Camera?

YouTube video by Every Frame a Painting

www.youtube.com

G_Dino @gprdino.bsky.social · Jan 19

Mofo's got jokes.

#Gemini #LLMs #GenAI

G_Dino @gprdino.bsky.social · Jan 15

Very nice writeup on #GestaltPrinciples in #design, particular #UI

#UXdesign #UIdesign

medium.muz.li/gestalt-prin...

Gestalt principles in UI design.

How to become a master manipulator of Visual Communication.

medium.muz.li

1

G_Dino @gprdino.bsky.social · Jan 14

More system 1, less system 2 — with all the good, bad, & ugly that implies. Younger folk are, as always, more affected.

If AI tools are causing you to offload instead of redirect #CognitiveLoad, you're happily walking into a fluency trap.

#CognitiveOffloading #AI #CriticalThinking

Increased AI use linked to eroding critical thinking skills

A study by Michael Gerlich at SBS Swiss Business School has found that increased reliance on artificial intelligence (AI) tools is linked to diminished critical thinking abilities. It points to cognit...

phys.org

Reposted by G_Dino

Prof Sam Illingworth @samillingworth.com · Jan 12

💉 Medical AI: Poisoned Data Risk

A new study highlights how deliberate misinformation can compromise medical AI models. Even 0.001% poisoned data increases harmful outputs.

Proposed solutions include real-time validation with knowledge graphs.

🔗 www.nature.com/articles/s41...

#SciComm #AI 🧪

Medical large language models are vulnerable to data-poisoning attacks - Nature Medicine

Large language models can be manipulated to generate misinformation by poisoning of a very small percentage of the data on which they are trained, but a harm mitigation strategy using biomedical knowl...

www.nature.com

3 34 86

G_Dino @gprdino.bsky.social · Jan 12

#Hallucination is one thing, but incorrect #retrieval multiplies the possible sources of where-the-fuckup-is. If it's not an #embedding, then it's a simple scrape. Is THAT the issue? I've seen screenshots of crisp text fail in OCR but if it's the same resource in PDF with strings, there's no issue.

G_Dino @gprdino.bsky.social · Jan 12

So who makes shit up less?

Google's Gemini Flash 2.0 ties with Chinese LLM Zhipu AI GLM-4 9B (chat), reportedly according to Vectara.

#GenAI #GenerativeAI #Gemini #LLM #Hallucination #OpenAI

Ranked: AI Models With the Lowest Hallucination Rates

Hallucination rate is the frequency that an LLM generates false or unsupported information in its outputs. Which models have the lowest rates?

www.visualcapitalist.com

2

G_Dino @gprdino.bsky.social · Jan 10

One one hand, I probably would have tried too, if I could. On the other........

No see, both my hands are free because it's an AI-powered turret. That's the point.

#OpenAI #AI #GunTurret?

OpenAI Shuts Down Developer Who Made AI-Powered Gun Turret

The DIY robot used ChatGPT's Realtime API to respond to commands to aim and fire a rifle.

gizmodo.com

1 3

G_Dino @gprdino.bsky.social · Jan 8

My mood these past few days:

The Moment by Saint Motel
All This Time by Skindred
Sleep Now in the Fire by Rage Against the Machine

lol

#SaintMotel #Skindred #RageAgainsttheMachine

G_Dino @gprdino.bsky.social · Jan 7

Hunch.tools is where it's at. #UI for #GenerativeAI pipelines, #AgenticOrchestration, even just single use cases (e.g. clean scraping or QC-ing / QA-ing repeatedly).

Made a "re-voicer" that restyles an artic from a URL into Pirate, Valley Girl, or I'm 14 and this is deep style. lol

Hunch | The AI-First Workspace Tool for Teams

Hunch is the AI-first workspace tool for teams to use all the top AI models, tackle complex projects, and be more productive.

Hunch.tools