AI is Getting Dumber. That's Not A Good Thing.

Excellent presepective

13:30 Is our society secretly run by the producers of the movie Idiocracy? Because that's what this sounds like.

The disingenuous examples of using old stagnant voice models or “instant” models to say AI isn’t getting smarter…. The frontier is on reasoning models thinking for longer and collaborating with each other. Voice models aren’t related to that or utilizing that at all and have been frozen in time since GPT-4, and instant models haven’t been much better

The only thing AI is becoming better at is coding sadly... even then, it needs someone who is educated in software development best practices to be piloted effectively.

When people come to the realization that AI is math, cannot think, isn't intelligent at all, and only predicts and calculate, then they'd understand why it behaves in the manners it does.

Gen, how are we supposed to trust your judgement/reasonings on AI if you've got a conflict of interest by being directly sponsored by an AI company?

💯.

It'd be good to clarify you're speaking about generative ai and not general purpose or specialized ai's as those are what i believe will actually make a difference, they already have but we aren't investing in that.

The reason, money, so when generative ai becomes a propaganda tool (i won't be surprised,) but there are actual good uses for general/specialized ai. It's already being used for shipping logistics, medical care, including cancer detection and treatment, doing invoicing and similar office work. Advancing scientific research, like more gmo's to protect endangered crops, better data analysis, the ability to track more variables at once and more.

If u prompt like an idiot or Write to an ai like to an other human, u will get such stupid Answers.... U have to prompt Specifically, precisely, because the AI doesn't think but calculates probabilities. And it's getting smarter all the time; without a doubt, AI is smarter than 80% of humanity. But that's only because 80% of the world Population is below average in intelligence.
The same applies to AI stealing ur jobs: it not steals your jobs, atleast not directly, but people who can use AI take your jobs. The stupid ones are now being weeded out of society.

The animations are giga distracting. Is your audience kids with adhd?

5:23 you're kidding 😒

"AI is so dumb"

Now onto our sponsor, AI

AI discussion in internet:
1- AI will become super intelligent and kill us all
2- AI is super dumb and useless.
And there is no third option what so ever.... 😂

the "how many letters ... " is a known limitation based on what unit tokens are.. stop ✋️ using that as dumbness measure.

sometimes you have to ride against the current or horizontally. the current is AI

well, too bad that the only thing AI needs, a genuine human creative thought, is something I will, by all means, try to prevent feeding to the AI since I'm sick of it existing in the first place.

AI can only exist by stealing every intellectual property from us. every idea. every visual art, every auditory one. why bother feeding the only thing that we have left to the AI? just so it can even replace that for us in a couple years?

in creative branches especially, AI is nothing but cancer. who doesn't say that if I use AI to potentially 'enhance' some of the ideas I have, that the owners of said AI won't just steal those ideas with impunity just like they already did before?

we live in very scary times. can't say I'm surprised though. how many months ago was it again when around 3 million files were made 'public'? how come those involved are still running free, huh?

since everyday we get more proof of the system being rigged, why trust the system at all.

AI isn't "getting dumber", it's just getting exposed for how dumb it always was. Press conferences and neat tech demos can only go so far once the product is in the hands of customers/businesses.

So..... it's emulating humanity's perfectly...... the newer the generation are, the dumber it gets...... so now it's at the Gen-Z era?

LoL it started eating it's own sh**

Says not to use ai to replace human thinking, advertises a tool which does exactly that.

"I'm not against AI, but what I'm heavily against is using AI for lazy slop and as a tool to replace human creativity" Finally someone with more than two brain cells !

More expect to always improve AI and their wealth, those CEO are greedy idoits

I’m always amused by people who lean heavily on AI in the workplace, then act surprised when their job is lost to automation. There’s even people paying for training, who don’t bother to actually study and just let AI take care of everything... then wonder why they’re unemployable.

AI needs to devalue soon, I need a new SSD.

There is "round 1", "round 2" in your video, meaning that somebody deliberately made AI to admit 2+2 is 5 by inventing a game

So that is basically a counterargument - people desperately trying to prove AI is wrong.

Don't get me wrong I also think what is happening is not right, just stating the obvious.

I am using Gemini for research to push me to research those topics that I don't know or understand, but I know Gemini is the best AI for it

A lot of it is that its being specialized for coding. So they get better at programming but worse at everything else. Coding is a huge chunk of what these are used for.

🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡

I fucking hate AI and just don't use it.

0:27 so, it's one of the stupidest examples I've ever seen in a YouTube video to demoralize AI. Listen buddy, did you ever heard about agentic AI, open globe, this type of system? You know, if a AI system have to be super smart and accurate, you know, it is multiple AIs. First of all, whatever this person have learned. It's a specific AI. It's a language model. It's not a tool or timer which it has access to calculate how many times you have gone. He doesn't have video access. He doesn't have audio access. Did you have on the camera or what? No. So we need more tools which will be add on after combining all the tools like the AI and I'm saying the future AI would be a system of all specific domains. LLM is a language model and tools will be different as image detection tool. Your iPhone's camera will detect you whether you are present or not. Your location will detect you where are you and your everything which will be fully calculated and then you will see the real power of AI. He will say brother, you are not moving, why are you not moving? Why should I count? It would sell like this. So please don't take specific example by just using free model and everything. Future AI is all about tools and all the extra domains it would combine. That's why Sam Altman is so crazy about a device specifically for AI device, openAI

can i ask who seriously reads this title and proceeds to click on the video to watch it with full seriousness?
are you real people?

We need to stop building huge context models

Consciousness is like philosopher's stone. It seems achievable, but all it will remain is an illusion.

It's all very simple garbage in => garbage out. Garbage made out of garbage is still garbage. If you want PHD level then you need to pay the $$$ and take the time to feed it PHD level data and training. There's a time and a place for AI. There's a lot of possibilities for what it can do. The problem is that the heads of these major companies are too disconnected from reality to understand even these basic things and maybe also don't really care about the product as long as they are getting rich.

is for me, I make good money doing the silly training gig more ez work pls lmao

___ing hilarious, your setup for the pro AI ad read. Ah, a few thousand bucks a pop must feel great, eh?

Please stop saying AI generally instead of saying LLMs...

Short AI stocks!

I use deepseek v4 und it so incredibly strong, much stronger than v3. I have a good process set up and I feel it is often times outsmarting me

In the last 2-3 months I have noticed Grok, Perplexity, and ChatGPT answers have become significantly less accurate.

You lost me with "models are getting dumber while companies tell you they're getting better." You can't say that with a straight face. Nobody thinks GPT-2 is smarter than what we have now. That claim is just wrong, and when it's the thesis of the whole video it makes the rest a lot harder to take at face value.

You also frame "less new data to gather" as if it means AI can't reach what a PHD knows. However, there's obviously enough on the internet for a person to earn a doctorate.

You're right that models are agreeable, that they hallucinate, and that they have real weaknesses humans don't though, but that wasn't your only claim.