Are Large Language Models Really AI?

Companies who operate call centeres:

They refuse to hire people in the nations they operate in, and instead setup call centeres in places like India cause they can pay them 10 cents a day there (they can absolautely afford to setup these things in the home nations, they just have investors demands to continually increase profits so they do this scummy stuff).

This allows these people who work there to get expereince interacting with western populations, they then move on from working for 10 cents a day and setup scam operations to continually phone western numbers to scam money out of your granny instead.

The 2nd thing would not exist if western companies were not taking advantage of cheap labor in the 3rd world in the first place, so in effect the call centers for your ISP, Bank etc and other companies are causing the scammers/theft we all have to put up with.

Also in relation to AI, these companies are now using AI products that change the voice of the person you are talking so in real time. So it can be an Indian man and they can make it sound like a western man.

No doubt in a few years, the scammers will also be using this AI tech too and it will be harder for "this is your bank sir" scammers to be instantly detected.

Radial that's just how modern capitalism works.

No in a few years about it. This is happening today.

Also:

Scammers Furious That Their Fellow Criminals Are Using AI, Saying It’s Unethical

No. Freaking. Comment. Just giggles.

I just heard about a news, that if its true, its good.

The news context is that Elon Musk had a large infrastructure, Colossus, to host her AI business, xAI.

BUT... xAI is behind and is only using 11% of the infra provided.

Giving that Elon Musk need results to make xAI public, the news is that he is renting Colossus to Anthropic. So that he can earn revenue and screw even more with Sam Altman.

There is this talk that Claude limits can go up.

IDK, I tried this weekend and didnt see anything different. People are saying changes are affecting only the busy hours.

We will see

The billionaires are like lions, when the investor money and market demand is unlimited they live with each other. When the investor money and users start slowing down, they start killing each other / buy out / demand more government regulation (only they can afford to follow) / do whatever until 1-2 mega companies remain.

Altman ain't gonna make it.

I think Claude will end up being eating by someone else, that could be Elon Musk the way things are headed.

Microsoft are floundering pretty hard lately.

Would this possibly be the same Elon Musk that promised actual full self driving cars in six months every year for TWELVE YEARS?

I'm cheering for Elon to hit 1 trillion net worth just for the comedy, don't feel strongly one way or another about most of his businesses or products.

I find it fascinating how willingly everyone goes along with the financial system right now just to watch number go up.

He also inspired RMT how to name their galkas after baby Aklklsdjfjk63476fjkd was born

Decent onformative video to catch up a little

IDK if it is impression, but this weekend my weekly quota feels like taking longer to fill.

As the usage page only gives percentage, maybe they ninja added extra tokens for model usage.

5h windows feels like about the same.

this is the way

wonderful reminder these things only have as much power as you give them, thanks for sharing!

While not truly intelligent AI is smarter than we think.

Being a Crappy Boss to AI Chatbots Pushes Them Toward Spouting Marxist Rhetoric and Organizing With Their Compatriots, Researchers Find

"Without collective voice, 'merit' becomes whatever management says it is."

I saw a thing last week where companies are intentionally lying about AI now.

They are saying things are AI, when it's literally just some guy remote controlling a thing from India. Security guard things, assembly line "robots", and many other things.

fake it til you make it

Always enjoyed a failed AI integration with robotics.

Opus 4.8 just released.

Benchmark and blablabla, but what feels nice is that they mentioned the 4.7 weak points, and this version tried to fix those.

Oh don't worry there will be new improved weak points.

I now assume everyone online is a bot connected to ChatGPT. Every social media site is now flooded with AI Bots spamming BS.

I'm so glad I quit social media in time to beat the rush.

These Are The States Where You're Stupidest And Smelliest, According To ChatGPT

In new research, the AI model had wild answers about where the sexiest and ugliest Americans live, too.

TLDR: We all have biases. AI internalizes them.

At least now you are putting the bare minimum effort when posting.

The first safe was impressive but after the fall he just gave up. The off stage drag must have been so painful.

GIGO, Garbage In Garbage Out. The first attempts at LLM's was just feeding them straight statistics and the results were hilarious.

I never post in the nude.

My spyware says otherwise

Employees Are Using Their Jobs’ Super-Expensive AI Tokens for the Most Hilariously Pointless Tasks Imaginable

Or, tokenmaxing, what is it good for? Also, why?

Hey, if Chanti puts in the bare minimum, it's the best damn minimum out there.

I just heard a good passage.

Any thing that is worth managing, it must be measure before being able to control it.

But sometimes, they just pick a metric thats "easy" to measure instead of picking a metric that actually measure what you want to reduce/improve/increase.

In theory, you would need to measure net gain in productivity with AI to determine the effectiveness or no of that tool.

But it is not easy to measure that, as it is a new thing, there isnt a consensus about the "best" way to use it, etc.

So, managers go with the "easiest" metric to measure AI usage: just measure tokens used. It MUST be related to performance, right?

Definitively no, it just made people to consume pointless tokens just for some bonus on "productivity".

But with the bills exploding, now the managers are reconsidering this metric.

My company, so far, doesnt have any metric to measure AI usage. And we are going to a very restricting policy to use tokens from external models in Copilot (somehow, we have house-hosted models, weaker but "free" to use)