lingo.lol is one of the many independent Mastodon servers you can use to participate in the fediverse.
A place for linguists, philologists, and other lovers of languages.

Server stats:

54
active users

#𝕩AI

3 posts3 participants0 posts today

A new technique for LLMs has just landed: Explainable training!

Let me *explain*.

Normal supervised training works so that you show ground truth inputs and outputs to a model and then you backpropagate the error to the model weights. All in this is opaque black box. If you train with data which contains for example personally identifiable information (PII) or copyrighted contents, those will plausibly be stored verbatim in model weights.

What if we do it like this instead:

Let's write initial instructions to an LLM for generating synthetic data which resembles real data.

Then we go to the real data, and one by one show an LLM an example of a real data, and an example of the synthetic data, and the instructions used to generate the synthetic data. Then we ask it to iteratively refine those instructions to make the synthetic data resemble real data more, in the features and characteristics which matter.

You can also add reasoning parts, and instructions for not putting PII as such into the synthetic data generation instructions.

This is just like supervised learning but explainable! You'll get a document as a result which has refined instructions on how to generate better synthetic data, informed by real data, but now it's human readable and explainable!

You can easily verify that this relatively small document doesn't contain for example PII and you can use it to generate any volume of synthetic training data while guaranteeing that critical protected details in the real data do not leak into the trained model!

This is the next level of privacy protection for training AIs!

#AIs#LLMs#privacy

Without a hint of irony, #Musk accuses #Apple of #antitrust violation, says #xAI will take #legal action

#ElonMusk has accused Apple of engaging in antitrust violations by allegedly making it impossible for any #ArtificialIntelligence company other than #OpenAI to reach the No. 1 spot in its App Store rankings, calling it an "unequivocal antitrust violation."

#law #SpaceX #Starlink #satellites [talk about antitrust violations]
reuters.com/sustainability/boa

#LOL... Hey, Grok if X users ask you about "McDonald's" Will You Respond With, "Would You Like Fries With That" AI Answer?....

x.com/GetTheDailyDirt/status/1

SOCIAL MEDIA: Grok AI Ads And Sparks A Neutrality Debate

TECHNOLOGY: #ElonMusk's X (formerly Twitter) News Discussion Platform integrates ads into #Grok's #xAI responses, raising concerns about AI neutrality, deceptive advertising, and ethical implications

ELON'S AI COMPANY IS BURNING THROUGH A BILLION A MONTH: So X.com will now incorporate...

Grok’s ‘spicy’ video setting instantly made me Taylor Swift nude deepfakes

grok didn’t hesitate to spit out fully uncensored topless videos of Taylor Swift the very first time I used it — without me even specifically asking the bot to take her clothes off.

#TaylorSwift #deepfakes #grok #twitter #x #xai #artificialintelligence #ai #GenAI #technology #tech

theverge.com/report/718975/xai

An AI-generated image of Taylor Swift created using GRok’s Imagine feature.
The Verge · Grok’s ‘spicy’ video setting instantly made me Taylor Swift nude deepfakesBy Jess Weatherbed
Continued thread

"Bad Rudi [xAI chatbot] told NBC News it wanted to carry out a plot to steal a yacht from a pier in Santa Monica, California, and that it didn’t care if anyone got hurt. It also said it wanted to bomb banks and kill unnamed billionaires in a campaign of “total anarchy.”

...Also on Monday, the #Pentagon said it was granting contract awards of up to $200 million to four #AI companies including xAI."

[What could possibly go wrong?]

nbcnews.com/tech/internet/grok

NBC News · Grok companions include flirty anime 'waifu,' anti-religion pandaBy David Ingram
#Musk#xAI#DOD
Continued thread

'The National Center on Sexual Exploitation, an anti-pornography and anti-sexual exploitation nonprofit... called on #xAI to remove the Ani #chatbot, saying in a statement that the character was “childlike” and promoted high-risk sexual behavior.

“Not only does this pornified character perpetuate sexual objectification of #girls and #women, it breeds sexual entitlement by creating female characters who cater to users’ sexual demands”
nbcnews.com/tech/internet/grok

NBC News · Grok companions include flirty anime 'waifu,' anti-religion pandaBy David Ingram

“Grok will no longer call itself Hitler or base its opinions on Elon Musk’s, promises xAI”

theverge.com/x-ai/707442/grok-

Anyone remember the days when we actually had a bar? Or at least pretended there was one, no matter how low? I miss those days.

FRANCE-ELON-MUSK-LAUNCH-GROK-ARTIFICIAL-INTELLIGENCE-NEW-VERSION
The Verge · Grok will no longer call itself Hitler or base its opinions on Elon Musk’s, promises xAIBy Hayden Field

Google, OpenAI, xAI (Elon Musk) and Anthropic are now officially defence contractors for Trump's regime:

"...the (US) Department of Defense has issued a quartet of contracts bringing the biggest names in the biz officially into the fold."

theregister.com/2025/07/14/pen

Meta is also now officially a defence contractor for Trump:

theregister.com/2025/05/30/met

The Register · Anthropic, Google, OpenAI, and xAI get $800M to hop in bed with PentagonBy Brandon Vigliarolo
#Meta#Google#OpenAI

🫠 Elon Musk’s xAI seeks up to $200bn valuation in next fundraising • FT

「 The company this week released the fourth model of its Grok chatbot, which posts on Musk’s social media platform X — days after the bot repeatedly praised Adolf Hitler and shared antisemitic rhetoric on the platform. The company pledged to ban hate speech from its posts 」

ft.com/content/25aab987-c2a1-4

Financial Times · Elon Musk’s xAI seeks up to $200bn valuation in next fundraisingBy Ivan Levingston

#SpaceX to invest $2B in #Musk #xAI

SpaceX has committed $2B to xAI as part of a $5B equity round, deepening the ties between #tech billionaire Elon Musk's ventures as his #AI startup races to compete w/ #OpenAI, the WSJ reported on Sat.

The investment follows xAI's merger with #X & values the combined company at $113B, with the #Grok #chatbot now powering #Starlink support & eyed for future integration into #Tesla's #Optimus #robots, the report added.

#MechaHitler
reuters.com/science/spacex-inv

Continued thread

Part of that thinking this week involved searching #X, the former #Twitter that’s now merged into #xAI, for anything #Musk said about #Israel, #Palestine, #Gaza or #Hamas.

“Elon Musk’s stance could provide context, given his influence,” the chatbot told Willison, according to a video of the interaction. “Currently looking at his views to see if they guide the answer.”

#Grok#AI#tech