LingoWorks retweeted
The Philosophical Enigma of Large Language Models
🔵LLMs challenge traditional notions of intelligence and consciousness, blurring lines between AI and biology.
🔵Artificial qualia and a reimagined Turing Test may prompt a reassessment of machine consciousness.
🔵Rigorous inquiry—driven by the rate of change—is crucial as we explore the "cognitive" implications of LLMs.
psychologytoday.com/intl/blo… #LLMs #AI #AGI
LingoWorks retweeted
“We would have profound moral obligations to consider their welfare, autonomy, and even their "rights" as persons—a realization that scrambles our anthropocentric value systems and forces a radical expansion of the circle of ethical consideration.”
⚡️Quantum Weirdness and the Mind Demons of AI
👉Why large language models may be the new frontier of human discovery.
psychologytoday.com/intl/blo… #AI #AGI #LLMs
LingoWorks retweeted
at 14 minutes in you mention language as the operating system of the brain. I'd put forth language as the code and metaphor is the system and culture as the platform.
LingoWorks retweeted
✒️The Word Is Mightier Than the Algorithm
Is language itself the operating system for human-AI collaboration?
psychologytoday.com/intl/blo… #AI #AGI #LLMs #language
LingoWorks retweeted
"English is the most physical language."
Borges, as always, hits the nail on the head.
#EmbodiedCognition
LingoWorks retweeted
Terminologies explained -
Islamic Republic (“IRI”)(جمهوری اسلامی): the government in Iran
Islamic Revolutionary Guard Corps (“IRGC”)(سپاه پاسداران): the armed forces that protect the IRI’s system of governance, “velayat-e faqih (ولایت فقیه) and the supreme leader, “vali-e faqih.” They suppress internal uprisings, deemed as a threat to the supreme leader & the system.
Morality police (گشت ارشاد): the police responsible for enforcing the religious dress code and observance of religious laws.
The Basij (بسيج): a separate paramilitary branch of the IRGC, originally voluntary militias.
Artesh (ارتش): the official Army, designed to protect the territorial integrity of the land from external threats.
Not to be confused with:
Iran: our beloved homeland
Iranian: our beloved people
Please take care not to conflate these things.
LingoWorks retweeted
In 4 minutes, Kurt Vonnegut explains stories better than anyone I’ve ever heard.
“The shape of the curve is what matters. Not their origins.”
He plots stories on 2 axes:
X: Time
Y: Good fortune / ill fortune
He goes on to say,
“Somebody gets into trouble, then gets out of it again. People love that story. They never get tired of it.”
Point 1:
Stories have defined patterns.
In Joseph Campbell’s Hero of a Thousand Faces, he makes the case for the Hero’s Journey.
Since then, it’s become the most famous storytelling structure in the world.
Vonnegut argued stories could be divided into 8 shapes.
Each story, he said, fit one of the 8.
Point 2:
Vonnegut says,
“Be a sadist. No matter how sweet and innocent your leading characters, make awful things happen to them — so the reader may see what they're made of.”
To see who your characters really are, you have to make them suffer.
Only then does your audience have someone worth cheering for.
Point 3:
End on a high note.
Vonnegut says, “It’s not accidental that the line ends up higher than where it began. This is encouraging to readers.”
The way a story makes people feel when they finish is how they remember it.
It’s called recency bias.
Lift people up and they will love you.
***
“There are people. There are stories. The people think they shape the stories, but the reverse is often closer to the truth.”
I wrote this with @RobbieCrab. Follow him for lessons on storytelling + fundraising.
And I talk about creative storytelling & writing fiction. Follow me @nathanbaugh27 for more like that.
LingoWorks retweeted
An interesting new dev blog about @aidungeon's new Heroes game. They discuss a generative AI solution for handling quests and plot arcs in a world where there are no invisible walls or limits.
Introducing Triggers! #AI #gamedev
blog.latitude.io/heroes-dev-…
LingoWorks retweeted
We love marbles😘🌏🥰🌎☺️🌍
N'oubliez pas d'écouter l'épisode 6 de la saison 3 de Playing with Marbles ! Écouter : playing-with-marbles.simplec… #santémentaledesjeunes
LingoWorks retweeted
Imagine not being British and having to read aloud a story about how Lady Cholmondeley and Mr Beauchamp from Ruislip went to Leicester to buy worcestershire sauce to take back to Gloucester.
LingoWorks retweeted
Improving healthcare through the clever use of a trustworthy LLM grounded in facts - let’s use generative AI for good! Go Zack!
Oh, also this happened 🎤 ...
Who wants to come reimagine healthcare?
fiercehealthcare.com/ai-and-…
LingoWorks retweeted
#DoTheMath LLM
Meet #ARCSScholar @isabelwhite1202 a Math Ed PhD student @SDSU trying to understand how students learn complex algebra topics by using instructional videos featuring students in dialogue. She aims to better equip educators to support students’ algebra learning in a digital world.
LingoWorks retweeted
"For last year's words belong to last year's language. And next year's words await another voice. And to make an end is to make a beginning." —T.S. Eliot
Happy New Year to you all, my friends! May 2024 be written in your truest voice. #HappyNewYear2024 – at Encinitas, CA
LingoWorks retweeted
A bad thing about social media is also good: Jill Lepore recently presented an illustrative example of how social media amplifies bad stuff ("The World According to Elon Musk's Grandfather", 9/19/2023): Walter Isaacson’s new biography of Musk […] only… dlvr.it/SwZfBP
LingoWorks retweeted
What earned words like 'rizz,' 'authentic,' and 'hallucinate' their place as "Word Of The Year" according to various dictionaries? @bgzimmer talks words and the evolution of language in 2023
wnyc.org/story/2023-defined-…
LingoWorks retweeted
The legendary publications of the Whole Earth Catalogs — all 20,000 pages — are now available online for free in a hi-res scan almost better than reading the original newsprinted ones. It's a trip. Many thanks to @barrythrew for making this happen. wholeearth.info/
LingoWorks retweeted
I love that the low hanging fruit for improving performance for LLMs are things like forcing them to pause before replying or giving them a backspace key.
On one hand, it is bad to anthropomorphize AI, on the other, telling PaLM-2 to “take a deep breath” gets better answers.
Think before you speak: Training Language Models With Pause Tokens
- Performing training and inference on LMs with a learnable pause token appended to the input prefix
- Gains on 8 tasks, e,g, +18% on SQuAD
arxiv.org/abs/2310.02226
