Is the AI-bubble about to pop?

[Zvoni]

Right,

NVidia losing 580 Billion USD overnight on stock-value due to some claim from a chinese startup (DeepSeek)
https://www.forbes.com/sites/dereksa...es-ai-darling/

Funny, how D. Trumps announcement to invest 500 Billion USD in "AI-research" in the US correlates......
Now, where to source those 500 Bil. USD now..... should be a nice challenge for the secretary of finance, no?
Or maybe give the job to Elon.... decisions, decisions....

[sapator]

That is an excellent post.

"AI chip giant Nvidia and other tech firms connected to AI, including Microsoft and Google, saw their values tumble on Monday in the wake of DeepSeek's sudden rise."

Deep in the covers US is not fighting Russia or Europe, it's fighting China.
And that is an undeclared war that is going on.
Unfortunately your link shows a white page here so I take the liberty to post another:

https://www.bbc.com/news/articles/c0qw7z2v1pgo

[TysonLPrice]

That is trusting the Chinese are telling the truth. I saw some people that had a chance to test it say it is equivalent to the "big boys". But that means you are trusting what they say about the cost.

[Zvoni]

That is trusting the Chinese are telling the truth. I saw some people that had a chance to test it say it is equivalent to the "big boys". But that means you are trusting what they say about the cost.

Well, i did write

NVidia losing 580 Billion USD overnight on stock-value due to some claim from a chinese startup (DeepSeek)

and a "claim" is exactly that: A statement, which "lacks" corroboration/evidence

[sapator]

That is even worse , if a "claim" makes these companies lose billions.

[wqweto]

and a "claim" is exactly that: A statement, which "lacks" corroboration/evidence

https://github.com/huggingface/open-r1

Is a reproducable claim still only a claim?

cheers,
</wqw>

[Zvoni]

https://github.com/huggingface/open-r1

Is a reproducable claim still only a claim?

cheers,
</wqw>

If a claim is "reproducable", then it is evidence/corroboration in itself, and the claim becomes fact (like "bug-hunting" in Software, and that still depends if it is a real Bug, or if it is intended behaviour).

In our case: Are there other "Start ups", which can reproduce the methodology DeepSeek used (500K Chips / 5 Million USD), and get an equivalent result?
If no, then it stays a (unproven) claim

[TysonLPrice]

It won't answer about Tiananmen Square or go against stated Chinese policy

What's next, they were the first to land on the moon

Then again...look what Google did:

Google Maps says U.S. users will see Trump’s new name for Gulf of Mexico:

https://www.washingtonpost.com/techn...le-maps-trump/

[sapator]

AI talk ....
I have a feeling that Moti will pop up soon enough. With the nickname MR Head.

[Zvoni]

Ah....well..

Let's wait for the first AI-Start-up Company having the balls to call themselves "Skynet".....

[sapator]

I would have gone with the John Connor foundation but I guess that will work also.

[TysonLPrice]

Ah....well..

Let's wait for the first AI-Start-up Company having the balls to call themselves "Skynet".....

I like HAL, but that dates me.

[sapator]

Oh come on!
Unfair!

[wes4dbt]

Deep in the covers US is not fighting Russia or Europe, it's fighting China.

It's pretty well known that China is close to becoming the worlds dominate country. With 1.4 billion people and a very low wage compared to the US, the US can't compete in manufacturing. We've also been losing our technology advantage. So as long as the people keep buying what the Chinese government is selling, it wont be long before they are the dominate power in the world.

What they will do with that is the big question.

[OptionBase1]

'Tis but a scratch.

[Shaggy Hiker]

What will anyone do with any of this? Humanity has a long track record of using new technology for good things...and for bad. There are some good things coming out of this. We'll just have to see what the bad parts end up being. I think Skynet isn't too unreasonable. One possible solution to any problem we have ever noticed would be to eliminate humans. After all, if you don't have humans, then there is nobody to notice that there are problems. It seems unlikely that a competent AI would not notice that there is one common denominator to all issues. The two questions about that would be: 1) Would the AI be able to take any action? and 2) Would we even notice in time to note our own demise?

[sapator]

The first target of Skynet would be Trump
Hopefully the next one our president.
One would be bound to stop Skynet (Musk would accidentally invent Skynet trying AI projections to Mars colonization) so he must be eliminated, the other one, well, to much stupidity that would be a constant danger to enter into a conversion with Skynet in the form of "I know you are but what am I?" that will initially go for 36,4 days so Skynet may overload and explode.

[wes4dbt]

One possible solution to any problem we have ever noticed would be to eliminate humans.

We are still very caveman like in some ways. We dress a little better and live in better caves.

[Peter Porter]

AI talk ....
I have a feeling that Moti will pop up soon enough. With the nickname MR Head.

Or maybe use an account he created a month ago, which will quickly discover his own chatbot.

[Zvoni]

The two questions about that would be: 1) Would the AI be able to take any action? and 2) Would we even notice in time to note our own demise?

@1) Depends if the 3 laws of Isaac Asimov are hardcoded into its source-code
@2) A big NO

[sapator]

I think the easiest would be overpopulation hunger. We grow like Locusts, there won't be enough food and we will happily die.
That may mean that maybe some will survive and we will start all over again.
The second scenario is nuclear annihilation, easy, easy, a trained dog can do it.
The third that is postponed by now is the Woke agenda, men marry men women marry women so no children.
The fourth is Skynet
The fifth is a real fakevid pandemia
And last the unavoidable, sun coming close to earth. Now we may not by then find a way to pull it away at 88 miles per hour but if we find a way to push it to 88...

(Popular deaths that unfortunately where Woke BS, ozone layer, the ice melt and drown us and of course climate change or climate crisis or however Wokers wanna call it)

[Shaggy Hiker]

The problem with Asimov's three laws is that ALL of his stories that involved the three laws were about how the laws were circumvented.

[sapator]

Elon Musk reportedly offers $97.4 billion for OpenAI, Sam Altman says 'no thank you'

https://www.yahoo.com/finance/news/e...215634193.html

[Shaggy Hiker]

Yet another billionaire slap fight.