Menu
News
All News
Dungeons & Dragons
Level Up: Advanced 5th Edition
Pathfinder
Starfinder
Warhammer
2d20 System
Year Zero Engine
Industry News
Reviews
Dragon Reflections
Columns
Weekly Digests
Weekly News Digest
Freebies, Sales & Bundles
RPG Print News
RPG Crowdfunding News
Game Content
ENterplanetary DimENsions
Mythological Figures
Opinion
Worlds of Design
Peregrine's Next
RPG Evolution
Other Columns
From the Freelancing Frontline
Monster ENcyclopedia
WotC/TSR Alumni Look Back
4 Hours w/RSD (Ryan Dancey)
The Road to 3E (Jonathan Tweet)
Greenwood's Realms (Ed Greenwood)
Drawmij's TSR (Jim Ward)
Community
Forums & Topics
Forum List
Latest Posts
Forum list
*Dungeons & Dragons
Level Up: Advanced 5th Edition
D&D Older Editions
*TTRPGs General
*Pathfinder & Starfinder
EN Publishing
*Geek Talk & Media
Search forums
Chat/Discord
Resources
Wiki
Pages
Latest activity
Media
New media
New comments
Search media
Downloads
Latest reviews
Search resources
EN Publishing
Store
EN5ider
Adventures in ZEITGEIST
Awfully Cheerful Engine
What's OLD is NEW
Judge Dredd & The Worlds Of 2000AD
War of the Burning Sky
Level Up: Advanced 5E
Events & Releases
Upcoming Events
Private Events
Featured Events
Socials!
Twitch
YouTube
Facebook (EN Publishing)
Facebook (EN World)
Twitter
Instagram
TikTok
Podcast
Features
Top 5 RPGs Compiled Charts 2004-Present
Adventure Game Industry Market Research Summary (RPGs) V1.0
Ryan Dancey: Acquiring TSR
Q&A With Gary Gygax
D&D Rules FAQs
TSR, WotC, & Paizo: A Comparative History
D&D Pronunciation Guide
Million Dollar TTRPG Kickstarters
Tabletop RPG Podcast Hall of Fame
Eric Noah's Unofficial D&D 3rd Edition News
D&D in the Mainstream
D&D & RPG History
About Morrus
Log in
Register
What's new
Search
Search
Search titles only
By:
Forums & Topics
Forum List
Latest Posts
Forum list
*Dungeons & Dragons
Level Up: Advanced 5th Edition
D&D Older Editions
*TTRPGs General
*Pathfinder & Starfinder
EN Publishing
*Geek Talk & Media
Search forums
Chat/Discord
Menu
Log in
Register
Install the app
Install
The
VOIDRUNNER'S CODEX
is coming! Explore new worlds, fight oppressive empires, fend off fearsome aliens, and wield deadly psionics with this comprehensive boxed set expansion for 5E and A5E!
Community
General Tabletop Discussion
*Dungeons & Dragons
Plagiarised D&D art
JavaScript is disabled. For a better experience, please enable JavaScript in your browser before proceeding.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an
alternative browser
.
Reply to thread
Message
<blockquote data-quote="EzekielRaiden" data-source="post: 9233505" data-attributes="member: 6790260"><p>It is not debatable. It is literally the nature of the computer programming required to achieve the ends they seek. You cannot improve a neural network's output unless it has fresh data to observe. That's how neural networks <em>work</em>. It's (effectively) a gigantic matrix of equations, and the equations can only be tweaked by feeding new, fresh, not-previously-observed data into them. It's a well-known problem in "AI" research right now--particularly because the demand for high-quality real-world data is expected to completely outstrip our supply of it within the next couple of years, and to even outstrip the supply of low-quality data within the next few decades.</p><p></p><p>An "AI"--a neural network--does not think. It does not decide. It literally can't. It is purely a statistical model--if you will, an extremely, mind-bogglingly advanced set of dice. It takes an input, jiggles it a little (likely through Brownian noise), and then passes that through its numerical matrix, weighting the different inputs in a bazillion different ways to then spit out an output (an image, a text, a piece of music, an analysis, whatever the network was trained to do.) At no point does it ever contain any amount of <em>meaning</em> or <em>symbolism</em>. It simply recognizes that character string "98290dFLsd890DF@#$" relates to character string "^&sdf23r:FPIOF)2q3PNVZ," which then relates to some other set of numbers, which then relates to some other set of numbers, (repeat this process dozens to hundreds of times), which then relates to some final set of numbers that are formatted to be read as the desired file output (image, text, sound, etc.)</p><p></p><p>The only way to improve a neural network that is based on something artistic--that is, art, music, literature, etc.--is to actually feed it examples that people liked and wanted to see. You <em>can't</em> create synthetic data of that kind. It's like trying to have a 5th grader teach herself algebra solely by having her write new math equations of the kind that she already knows--she'll never generate a <em>variable</em> because she's never been taught to. To say nothing of something like calculus!</p><p></p><p>Further, all that stuff you said about AIs "inventing a language"? Hogwash, mostly caused by media outlets getting a scoop without understanding. Here is <a href="https://towardsdatascience.com/the-truth-behind-facebook-ai-inventing-a-new-language-37c5d680e5a7" target="_blank">an actually sober explanation</a> of what happened with the AI "inventing its own language." TL;DR: The output was mostly gibberish and resulted from the AIs involved having a faulty reward mechanism, which judged only the negotiation result, not the way the negotiation was done--so if saying "I want" made a negotiation succeed more, then saying "I want" <em>twice</em> would totally make it succeed twice as much, right??? That's exactly the "thinking" that went into these bots' behavior.</p><p></p><p>You can also read, <a href="https://www.snopes.com/fact-check/facebook-ai-developed-own-language/" target="_blank">here on Snopes</a>, an actual statement from one of the researchers who worked on this project: </p><p></p><p>But, as usual, science literacy in journalism is rare to nonexistent, and thus we got ridiculous, alarmist "AI INVENTING NEW LANGUAGES!!!" It isn't. At the absolute most sophisticated, it is developing <em>encodings</em> which map similar structures to similar places in a massive poly-dimensional space.</p></blockquote><p></p>
[QUOTE="EzekielRaiden, post: 9233505, member: 6790260"] It is not debatable. It is literally the nature of the computer programming required to achieve the ends they seek. You cannot improve a neural network's output unless it has fresh data to observe. That's how neural networks [I]work[/I]. It's (effectively) a gigantic matrix of equations, and the equations can only be tweaked by feeding new, fresh, not-previously-observed data into them. It's a well-known problem in "AI" research right now--particularly because the demand for high-quality real-world data is expected to completely outstrip our supply of it within the next couple of years, and to even outstrip the supply of low-quality data within the next few decades. An "AI"--a neural network--does not think. It does not decide. It literally can't. It is purely a statistical model--if you will, an extremely, mind-bogglingly advanced set of dice. It takes an input, jiggles it a little (likely through Brownian noise), and then passes that through its numerical matrix, weighting the different inputs in a bazillion different ways to then spit out an output (an image, a text, a piece of music, an analysis, whatever the network was trained to do.) At no point does it ever contain any amount of [I]meaning[/I] or [I]symbolism[/I]. It simply recognizes that character string "98290dFLsd890DF@#$" relates to character string "^&sdf23r:FPIOF)2q3PNVZ," which then relates to some other set of numbers, which then relates to some other set of numbers, (repeat this process dozens to hundreds of times), which then relates to some final set of numbers that are formatted to be read as the desired file output (image, text, sound, etc.) The only way to improve a neural network that is based on something artistic--that is, art, music, literature, etc.--is to actually feed it examples that people liked and wanted to see. You [I]can't[/I] create synthetic data of that kind. It's like trying to have a 5th grader teach herself algebra solely by having her write new math equations of the kind that she already knows--she'll never generate a [I]variable[/I] because she's never been taught to. To say nothing of something like calculus! Further, all that stuff you said about AIs "inventing a language"? Hogwash, mostly caused by media outlets getting a scoop without understanding. Here is [URL='https://towardsdatascience.com/the-truth-behind-facebook-ai-inventing-a-new-language-37c5d680e5a7']an actually sober explanation[/URL] of what happened with the AI "inventing its own language." TL;DR: The output was mostly gibberish and resulted from the AIs involved having a faulty reward mechanism, which judged only the negotiation result, not the way the negotiation was done--so if saying "I want" made a negotiation succeed more, then saying "I want" [I]twice[/I] would totally make it succeed twice as much, right??? That's exactly the "thinking" that went into these bots' behavior. You can also read, [URL='https://www.snopes.com/fact-check/facebook-ai-developed-own-language/']here on Snopes[/URL], an actual statement from one of the researchers who worked on this project: But, as usual, science literacy in journalism is rare to nonexistent, and thus we got ridiculous, alarmist "AI INVENTING NEW LANGUAGES!!!" It isn't. At the absolute most sophisticated, it is developing [I]encodings[/I] which map similar structures to similar places in a massive poly-dimensional space. [/QUOTE]
Insert quotes…
Verification
Post reply
Community
General Tabletop Discussion
*Dungeons & Dragons
Plagiarised D&D art
Top