Author Topic: Deepseek and Qwen2.5 Max  (Read 833 times)

0 Members and 1 Guest are viewing this topic.

Offline tp4tissue

  • * Destiny Supporter
  • Thread Starter
  • Posts: 13684
  • Location: Official Geekhack Public Defender..
  • OmniExpert of: Rice, Top-Ramen, Ergodox, n Females
Deepseek and Qwen2.5 Max
« on: Wed, 29 January 2025, 10:08:04 »
So what's happening.

Our country's AI lead evaporated overnight.


Tp4 predicted this years ago
regarding technological convergence. This is a sobering moment for american AI companies.

American AI services as a commodity is basically destroyed.

American Cloud-Feudalism (Real Hardware), still remains world leading. If Tp4 were a gambler, puts on all AI model as a service companies (openai), calls on AWS, TSM, NVDA.

Offline tp4tissue

  • * Destiny Supporter
  • Thread Starter
  • Posts: 13684
  • Location: Official Geekhack Public Defender..
  • OmniExpert of: Rice, Top-Ramen, Ergodox, n Females
Re: Deepseek and Qwen2.5 Max
« Reply #1 on: Wed, 29 January 2025, 11:08:25 »
We're confident the JOB of writer is capoot.

This would take a decent writer at least half an hour.   Add unlimited context/referencing, no human writer can out write an LLM.

What intellectual domain isn't fundamentally rule-based?

_____________


312115-0

312117-1

Offline fohat.digs

  • * Elevated Elder
  • Posts: 6516
  • Location: 35°55'N, 83°53'W
  • weird funny old guy
Re: Deepseek and Qwen2.5 Max
« Reply #2 on: Wed, 29 January 2025, 11:15:49 »
Fascinating.
“He was very effective. He knows those computers better than anybody.
Those vote counting computers. And we ended up winning Pennsylvania like in a landslide.
So it was pretty good .... Thank you to Elon.”

- Donald Trump - Washington DC  2025-02-02

Offline pixelpusher

  • * Elevated Elder
  • Posts: 4217
  • Location: Tennessee - USA
Re: Deepseek and Qwen2.5 Max
« Reply #3 on: Wed, 29 January 2025, 11:24:17 »
We're confident the JOB of writer is capoot.

This would take a decent writer at least half an hour.   Add unlimited context/referencing, no human writer can out write an LLM.

What intellectual domain isn't fundamentally rule-based?

_____________


(Attachment Link)

(Attachment Link)

It is interesting.  Isn't the LLM just going through what it's already been fed?  So the big difference is that a LLM can never come up with an original thought.  It's only giving you what it's read before.  It's like a fast thought thesaurus.  Perhaps that's all we are as well though 🤔

Offline tp4tissue

  • * Destiny Supporter
  • Thread Starter
  • Posts: 13684
  • Location: Official Geekhack Public Defender..
  • OmniExpert of: Rice, Top-Ramen, Ergodox, n Females
Re: Deepseek and Qwen2.5 Max
« Reply #4 on: Wed, 29 January 2025, 11:35:47 »
It is interesting.  Isn't the LLM just going through what it's already been fed?  So the big difference is that a LLM can never come up with an original thought.  It's only giving you what it's read before.  It's like a fast thought thesaurus.  Perhaps that's all we are as well though 🤔

We can't answer foundational questions such as this because the answer is a frame of the whole.

My point is only that no writer can produce writing at a pace that would be economically competitive.

Offline fohat.digs

  • * Elevated Elder
  • Posts: 6516
  • Location: 35°55'N, 83°53'W
  • weird funny old guy
Re: Deepseek and Qwen2.5 Max
« Reply #5 on: Wed, 29 January 2025, 12:16:54 »

can never come up with an original thought.


can produce writing at a pace


What you receive, as I see it, is a regurgitation of "the opinion" of "the internet as a whole"

As long as living human beings are remembering that all they are seeing is an opinion we will be alright.
Once people attribute more value or weight to it than they would to the opinion of any other unreliable author, then we will be in trouble.
 
“He was very effective. He knows those computers better than anybody.
Those vote counting computers. And we ended up winning Pennsylvania like in a landslide.
So it was pretty good .... Thank you to Elon.”

- Donald Trump - Washington DC  2025-02-02

Offline tp4tissue

  • * Destiny Supporter
  • Thread Starter
  • Posts: 13684
  • Location: Official Geekhack Public Defender..
  • OmniExpert of: Rice, Top-Ramen, Ergodox, n Females
Re: Deepseek and Qwen2.5 Max
« Reply #6 on: Wed, 29 January 2025, 12:19:12 »
If climate change does not kills us, which it probably will.

AI's growth is unlimited, its competence already surpasses the average human, Sooner rather than Later, it will surpass even extraordinary humans, and finally, All humans.

Offline LavenderB

  • Posts: 41
  • Location: Poland
  • I believe in unix control supremacy
Re: Deepseek and Qwen2.5 Max
« Reply #7 on: Wed, 29 January 2025, 14:11:15 »

My point is only that no writer can produce writing at a pace that would be economically competitive.

Well, no writer can produce annoying corporate word soup at a competitive pace.
The actual articles and books that LLMs spit out are pretty terrible. Companies will obviously still want to automate everything because of greed, but I'm guessing that actual journalists and good writers won't be replaced in the foreseeable future.
What I'm mainly worried about is AI management or HR. We have to make it very clear that this would be unacceptable.
Safa | 1993 UK ISO-DE Model M  | Frog TKL | 3d printed bakeneko | Akko 5075s