From Trace Dominguez. ( YouTube / Nebula )
Half of Americans TRUST what AI says to them, but they should NOT. ๐
โโ๏ธ Feat. @JordanHarrod !
๐ Watch an ad-free, director’s cut version of this video on Nebula: http://go.nebula.tv/tracedominguez (Joining directly supports the channel!)
โญ โญ โญ S U G G E S T โญ โญ โญ
Want me to make a video about a burning question you have? A cool story you found? Think of me as your nerdy best friend. I’ll do the research and you get the benefits!
Suggest a video here: https://noteforms.com/forms/think-of-me-as-your-nerdy-best-friend-ezzqc3
๐๐๐ A U D I O ๐๐๐
ENGLISH
This video has been dubbed using an artificial voice. You can change the audio track language in the Settings menu. https://aloud.area120.google.com
SPANISH
Este video ha sido doblado al espaรฑol con voz artificial. Puede cambiar el idioma de la pista de audio en el menรบ Configuraciรณn. https://aloud.area120.google.com
PORTUGESE
Este vรญdeo foi dublado para o portuguรชs usando uma voz artificial. Vocรช pode alterar o idioma do รกudio no menu Configuraรงรตes. https://aloud.area120.google.com
๐โโ๏ธ๐โโ๏ธ๐ D E T A I L ๐๐โโ๏ธ๐โโ๏ธ
All three below are TRUE:
1. #ChatGPT and other ChatBot AI is the hot technology rn.
2. These #LargeLanguageModels don’t know anything.
3. Humans are terrible lie detectors.
These three truthfacts make AI’s overwhelming confident "speech" a kryptonite to human thinking.
The #Dunning-KrugerEffect is a psychological tool to understand cognitive bias. It says when people learn a little about a complex topic, they are over confident in their understanding. #LargeLanguageModels present this exact behavior to audiences. Not only can LLMs like ChatGPT not fact check their replies, they also have no way of knowing what they don’t know!
On top of that, most humans don’t know how AI chatbots (like ChatGPT) work, meaning our own Dunning-Kruger bias comes into play as well. Tech like ChatGPT cannot fact check itself! Here’s what you need to know.
This Dunning-Kruger Effect is AI’s Biggest problem, and ours too.
๐๏ธ๐๏ธ๐๏ธ P O D C A S T ๐๏ธ๐๏ธ๐๏ธ
I’ve got a BRAND NEW PODCAST called That’s Absurd Please Elaborate. In each episode, my buddy Julian Huguet and I take silly questions from the audience and answer them with serious research.
Some of my favorites:
– How long would it take to make a LEGO replica of the sun?
– What’s the pointiest thing?
– Why aren’t humans born with adult sized hands and feet?
– Where is the internet?
We release episodes every other Thursday!
๐ Nebula: https://go.nebula.tv/thatsabsurdshow
Or if you prefer to listen OFF Nebula, go here:
๐ Apple Podcasts: https://podcasts.apple.com/us/podcast/thats-absurd-please-elaborate/id1680094699
๐ Spotify: https://open.spotify.com/show/6JOkmJBsVOuODHarxzmotZ?si=bbd7324faa1f4d7d
Or check Pod Link, they’ve got all the apps: https://pod.link/1680094699
๐๐๐ A U D I O ๐๐๐
This video has been dubbed using an artificial voice via https://aloud.area120.google.com to increase accessibility. You can change the audio track language in the Settings menu.
๐งฒ๐งฒ๐งฒ F O L L O W ๐งฒ๐งฒ๐งฒ
Threads ๐งต:: https://www.threads.net/@tracedominguez
BlueSky ๐:: https://bsky.app/profile/tracedominguez.com
Instagram ๐ท:: https://instagram.com/tracedominguez
Twitter โ:: https://twitter.com/tracedominguez
Patreon ๐:: http://patreon.com/tracedominguez
๐๐๐ C H A P T E R S ๐๐๐
00:00 There are no AIs that can Fact Check
00:44 ChatGPTโs Dunning-Kruger Problem
1:50 What is the Dunning-Kruger Effect?
6:33 How does this apply to AI & ChatGPT?
7:41 Before you ask ChatGPT to write things, you should know
9:40 ChatGPT cannot Fact Check
11:18 The main takeaway from this videoโฆ
13:03 โI dream of a better worldโ
14:20 I have a podcast!
๐๐๐ L E A R N ๐๐๐
Cornell University’s Preprint: Extracting Training Data from AI Large Language Models (LLMs)
https://arxiv.org/pdf/2012.07805.pdf
Humans Are Pretty Lousy Lie Detectors
https://www.scientificamerican.com/article/humans-are-pretty-lousy-lie-detectors/
Unskilled and Unaware of It: How Difficulties in Recognizing One’s Own
Incompetence Lead to Inflated Self-Assessments (actual PDF)
https://www.avaresearch.com/files/UnskilledAndUnawareOfIt.pdf
ChatGPT is Dumber Than You Think
https://www.theatlantic.com/technology/archive/2022/12/chatgpt-openai-artificial-intelligence-writing-ethics/672386/
Google Warns Staff about AI ChatBots
https://www.reuters.com/technology/google-one-ais-biggest-backers-warns-own-staff-about-chatbots-2023-06-15/
Studies show ChatGPT cheating is on the rise among students
https://www.businessinsider.com/teachers-caught-students-cheating-chatgpt-survey-shows-2023-2
Vintage Wallpaper image by Getty Images
Music by Epidemic Sound
Thanks for supporting my scicomm! Stay curious. I’ll see you in the future. ๐e