AIGuys

✌I like to deflate AI hype and talk about real research.
✍AI Book:
☛ Editor & Blogger @ AIGuys

[AIGuys @RealAIGuys]

One of the biggest breakthroughs in Geometric #AI.

Now we can actaully control the geometry of a model and make it robust towards a certain type of noise.

#AIresearch
https://t.co/nlbfh1NbBf

[AIGuys @RealAIGuys]

Use humans to train these systems and then replace humans with these systems.

[Subbarao Kambhampati (కంభంపాటి సుబ్బారావు) @rao2z]

If true, it seems like we don't need AGI to steal our purpose.. Fellow humans with IPO pressures can do the needful.. 🙄https://t.co/0Esic4DT0q

[AIGuys @RealAIGuys]

My friends are still as dissatisfied or even more with the introduction of AI in their work as they were before the #AI.

[AIGuys @RealAIGuys]

@rom1trs OS

[AIGuys @RealAIGuys]

Why Supervised Learning Has a Blind Spot

#AI #Artificialintelligence #AIresearch
https://t.co/NGfbn3jjQF

[AIGuys @RealAIGuys]

What If Robustness in AI Is Not a Regularization Problem—but a Geometry Problem?

[AIGuys @RealAIGuys]

Supervised Learning Has a Geometric Blind Spot

https://t.co/EU7de7yXUW

[AIGuys @RealAIGuys]

@Number7AI is changing how financial Saas is built. Doing LLM based OCR is so 2025, now it is time for actively identifying what's going wrong and connecting with the larger financial ecosystem.

#Fintech #financeact2026 #Finance #Accounting

[AIGuys @RealAIGuys]

The irony is so big that Anthropic is crying about data stealing while simulatnaeously stealing the data from the entire humanity.

I really wonder how do these people sleep at night.

Is morality only left for the common folks?

[Ricardo @Ric_RTP]

Palantir's CEO just exposed Sam Altman and Dario Amodei for robbing every Fortune 500 company.

Within two minutes, Alex Karp took the entire frontier AI industry apart on national television.

His exact words:

"Every single enterprise in this country, these people are LIVID. They are paying for tokens that create no value. These people are stealing the weights and alpha of my business."

He literally said the entire frontier AI business model is intellectual property extraction dressed up as a subscription.

Then he also destroyed the pricing model with a single question that Silicon Valley still refuses to answer:

"If it was so valuable, let's say I can make you $1 billion tomorrow. Wouldn't I say I'll make you $1 billion and I want 30 percent? Why are they charging for tokens if it's so valuable?"

That question breaks the industry.

If OpenAI and Anthropic's models truly delivered the productivity gains the labs claim, they would take equity or a share of the profit they generate. They would not sell access by the million tokens.

Token pricing is itself the CONFESSION that the product cannot produce reliable value at scale. If it did, they would price for the value. But they price for the compute because that is what they are actually selling.

Karp went even further...

He called the entire arrangement "a wealth tax that does not help the poor. It just punishes."

American businesses are transferring the alpha of their operations, meaning the workflows, the customer data, the strategy memos, the internal models that make them competitive, directly into the training pipelines of a handful of Silicon Valley labs. Once those labs retrain, the customer's own edge becomes the next enterprise product sold back to their competitors.

And the part the AI industry does not want anyone thinking about:

Every enterprise running its confidential documents, its customer conversations, and its financial models through a frontier model is potentially teaching that model HOW to replace them.

The vendor collects the token fee AND the compounding intelligence about that customer's business. That is the mechanism. And that is why Karp used the word "stealing."

He claims this is why every executive he meets is furious in private and silent in public. Nobody wants to be the CEO who called out the labs and then discovered their next competitor was built on their own leaked workflows.

The entire AI industry has been priced for perfection on one assumption:

That frontier labs produce durable, defensible value that justifies infinite compute spend.

But Karp just told us that the customers do not believe that assumption anymore. They believe they are being taxed without benefit, watched without consent, and copied without recourse.

The moment enterprises stop believing, the whole valuation stack shakes.

[AIGuys @RealAIGuys]

@GeminiApp is getting stuck all the time, it is horrendous to use, it can't write more than 10 pages in one go, switching between canvas and other modes and also changing the model mid chat is so bad. Also why can't I edit my previous chats like claude?

[AIGuys @RealAIGuys]

@jab I feel with all these subscription costs sky rocketting, people will soon move to localized model, especially with the new AMD and Nvidia hardware, DGX & Ryzen AI Halo.

[AIGuys @RealAIGuys]

@GeminiApp boasts massive context windows, but it struggle so hard to even write 20 pages in one go. Why that's the case, I've no idea. It again and again keep giving me 5-10 pages, no matter how I prompt it.

[AIGuys @RealAIGuys]

Sonnet 4.6 is now terrible, I do feel they purposely downgrade these models significantly.

#Anthropic

[AIGuys @RealAIGuys]

Anthropic shipped Fable 5 and Mythos 5 today. They’re the same model. Who gets which version depends entirely on who you are. That’s a genuinely new idea in AI deployment and it came apart at the seams within hours of launch.

#Fable5
https://t.co/jPrdONXvZB

[AIGuys @RealAIGuys]

I feel AI Labs are doing the IPhone thing, making older models bad so people shift to new ones.

[AIGuys @RealAIGuys]

All the Foundational model evals can be broken easily. We have nailed down input output mapping no matter the complexity.

So, what's the next eval, something that measure internal consistency of the weights or geometry of the model. This will be much harder to game.

[AIGuys @RealAIGuys]

What if every trick for model robustness is actually the exact same math? https://t.co/qA9HWrk8DT via @LinkedIn

[AIGuys @RealAIGuys]

Thanks for sharing this.

[Distill AI @DistillAI1]

This paper unifies robustness, domain adaptation, and invariance problems under a single geometric framework called the "matching principle" for learning nuisance-robust representations. #ml #ai
https://t.co/hhnWBAQ5TZ

[AIGuys @RealAIGuys]

@KurateOrg It is good, but there are few things it misses, mainly the overall pathways it opens. Asking the right question is one of the cornerstone of science.

[AIGuys @RealAIGuys]

Our paper "The Matching Principle: A Geometric Theory of Loss Functions for Nuisance-Robust Representation Learning" ranked #16 in cs.LG Preprints on @kurateorg! https://t.co/SgbkJpv1Vj