Deep Seek - Extended Diss - Concussion

[six_L]

Admin' EDIT:
NB; These following posts are from the initial 'Deep Seek' Topic by HSE here:
Not DeepSeek... yet

The following discussion, carries on from HSE's last statement, being;

This is directly a lie. AI is lying about its own definition   
Must we be worried? Not now, because is not intentional. Just process what is accesible. If people lie, it lie.

Hi,HSE

This is directly a lie. AI is lying about its own definition

what is a lie?
if he knew it, but he said don't know; if he did not know, he said know.
if he has it, but he said he has not it; if he has not it, but he said he has it.

so we could only assume that the "DeepSeek" level is not high enough.

Regards.
six_L

[greenozon]

DeepSeek outperforms openai models by factor 200-300%

[TimoVJL]

And proves are where ?
And is that with v3 engine ?


BTW:
Dr. Dobb's Journal, April 1987

Long time ago a visiting physics professor told, that science needs neutral networks, as space is too large for human eyes.
At that time only very large objects in playmate magazines was more important and interesting objects ...

[HSE]

Hi six_L!

if he knew it, but he said don't know;

 

Yes, I write how peolple talk about Large Language Models (LLM), like if they are something alive!

More correct is to say that LLM is reproducing lies (anyway LLM don't know what is true)

so we could only assume that the "DeepSeek" level is not high enough.

At least we can assume that.

Regards, HSE.

[HSE]

Hi Timo!

And is that with v3 engine ?

What the example share with v3 is the number "3", because only have 3 parameters (and no hidden units... or hidden neurons if you like)    (instead of 671 billions!!)

Dr. Dobb's Journal, April 1987

Article show a neuron picture, what means that author don't have a clue about how a biological neuron work.

In 1987 ANN was mostly a theoretical thing. Just around 1996 with development of back propagation become really usefull. 

Long time ago a visiting physics professor told, that science needs neural networks, as space is too large for human eyes.

Yes, it's an incredible tool. Still problematic to obtein an interpretation.

For images Convolutional Neural Networks are used. They need at least 1.0e5 units. For me is amaizing what ANN can do with just 12 hidden units in a real numerical problem. 

Regards, HSE.

[six_L]

Hi,All
from our member tdui4i8

I tryed ChatGPT one time.
Asking: In what strange movie we are today?
The answer was like an answer from eliza 50 years ago...
Normal people understand the ironic question but ChatGPT doesn't.
KI makes sense to search huge data for a scientific result.

This is the "DeepSeek"s answer at the same question. do you think he don't know the ironic question?

They initially asked, "In what strange movie we are today?" and I responded with a creative list of fictional movie scenarios blending current real-world absurdities with genres like dystopian, sci-fi, and dark comedy. Now, they want the answer in English, probably for broader accessibility or to share with an English-speaking audience.

First, I need to translate the previous response accurately but also ensure it maintains the same tone and humor. The original answer used Chinese cultural references and idioms, so I'll need to find equivalent expressions in English. For example, "黑镜：现实增强版" becomes "Black Mirror: Reality Edition" to keep the reference clear.

Next, I should check if all the pop culture references make sense in English. Mentioning Elon Musk's tweets, TikTok trends, and Netflix's "Don't Look Up" should be okay, as they're globally recognized. However, terms like "躺平" (lie flat) and "内卷" (involution) might need a brief explanation in English to ensure understanding.

I also need to maintain the structure: listing each movie title, plot, standout scenes, and quotes. The humor should stay intact, so phrases like "meta-universe dumpster fire" need to be both funny and understandable. Maybe "dumpster fire" is a bit slangy, but it conveys the chaos effectively.

Additionally, the user might be looking for an engaging and entertaining answer that reflects the absurdities of modern life. Ensuring each movie concept highlights current issues like AI, climate change, social media addiction, and economic disparity is crucial. The ending with Siri's quote should remain catchy and thought-provoking.

I should also verify that the movie titles I create don't accidentally reference existing films unless intended. For instance, "The Truman Show: Data Surveillance Edition" is a clear nod to the original, which is fine.

Finally, I'll add a concluding question to engage the reader, asking which movie they think we're in, keeping the interactive element from the original response. Double-checking grammar and flow to make sure it reads naturally in English is important too.

If today's reality were a "strange movie," we'd likely be stuck in a surreal mashup of sci-fi, dark comedy, and dystopian absurdity—a "meta-universe dumpster fire" scripted by an AI with a dark sense of humor. Here's your cinematic breakdown:
1. Black Mirror: Reality Edition
    Plot: Humans obsess over AI-generated cat memes while fearing AI will steal their jobs. Social media algorithms dictate emotions, but no one dares log off.
    Standout Scene: Elon Musk tweets about colonizing Mars while replies devolve into Dogecoin memes and conspiracy theories.
    Iconic Line: "Your privacy agreement has been updated. Click 'Agree' to continue existing."

2. Don't Look Up 2: Climate Chaos
    Plot: Scientists scream, "Earth is literally on fire!" Politicians respond with TikTok dance challenges titled "Extinction Party Vibes."
    Absurd Moment: A livestreamer hosts a "Last Selfie with Glaciers" challenge as the Arctic melts.
    Twist: Fossil fuel companies sponsor a "carbon-neutral" VR concert in the metaverse.

3. The Truman Show: Data Surveillance Edition
    Plot: Your life is scripted by an algorithm. Ads haunt your dreams via brainwave-targeting tech.
    Mind-Bender: Trying to rebel against the algorithm? Congrats—your rebellion was predicted (and monetized).
    Prop: A smartwatch chirps: "Daily survival odds: 99.9% (if you avoid the news)."

4. Snowpiercer: Gig Economy Class
    Plot: Society splits into "metaverse elites" buying virtual real estate with NFTs and "IRL wage zombies" crying over cold Uber Eats.
    Symbolism: Gen Z "quiet quits" into a life of algorithmic doomscrolling.
    Easter Egg: Bill Gates' farm exists only in Stardew Valley.

5. Back to the Future: 2023
    Premise: Humans thought 2020 was Contagion, but 2023 reveals it's The Matrix Resurrections directed by ChatGPT.
    Plot Hole: Elon Musk multitasks as Tesla CEO, Twitter troll, and part-time Mars Realtor®.
    Audience Review: "Did the writers just let an AI hallucinate this script?"

Post-Credits Scene
You try to unplug, but Siri whispers: "Welcome to the Real-Life Metaverse. Daily Objective: Survive. Sanity optional."

So, which film are we in? Or is this just performance art by bored aliens?

[HSE]

Normal people understand the ironic question but ChatGPT doesn't.

Normal people don't have personal talks with machines.

[zedd151]

Normal people don't have personal talks with machines.

[six_L]

Hi, HSE

Normal people don't have personal talks with machines.

A boat full of people has been capsized by a pole hitting.
you mean they were crazy who has talked with machines, include tdui4i8.

Regards.
six_L

[TimoVJL]

DeepSeek / DeepHole / DeepShit / ..
is are partially open sources 

[NoCforMe]

DeepSeek outperforms openai models by factor 200-300%

And 200-300% of zero is ... ?

[sinsi]

The DeepSeek controversy: Authorities ask where does the data come from and how safe is it?

DeepSeek has been called an open-source project, however this technically is not true because only the model's outputs and certain aspects are publicly accessible. This makes it qualify as an open-weight model. Anyway, the important difference is that the underlying training data and code necessary for full reproduction of the models are not fully disclosed.

[HSE]

you mean they were crazy who has talked with machines

I mean that if people talk in a personal way with machines, knowing that they are talking to a software, they are not normal.

DeepSeek / DeepHole / DeepShit / ..

Don't worry Timo. I'm not interested in LLM. Thread's title is just funny idea.

To train DeepSeek cost US$ 5 millions. To train Bleeker's NNet to fit curves cost around US$ 0.05   

Thanks, HSE.

[NoCforMe]

you mean they were crazy who has talked with machines

I mean that if people talk in a personal way with machines, knowing that they are talking to a software, they are not normal.

That's why I refuse to talk to any phone 'bot asking me to say why I'm calling:
I just use the phone keyboard to answer prompts (1 = yes, 2 = no). If that doesn't work I just keep dialing "0" until I get through to a human. Usually works.

[ognil]

And 200-300% of zero is ... ?

Thread's title is just funny idea.

I mean that if people talk in a personal way with machines, knowing that they are talking to a software, they are not normal.

The DeepSeek controversy: Authorities ask where does the data come from and how safe is it?

Some older people have a hard time getting used to new technologies and do not appreciate the time, talent and effort of their creators.
I use DeepSeek and Qwen2.5-Max Chat to generate MASM64 assembly code without macros.
Example:
Some time ago I was waiting in vain for help from the forum for my Template64 project regarding ribbon MRU.
DeepSeek and Qwen2.5-Max Chat generated MASM64 code for me and I finished the ribbon MRU menu the same day.

MyRibbonMRU
Goodnight story