Sam Altman is a habitual liar and can't be trusted, says former OpenAI board member

[Hood Critic]

Now we see the real reason he's gassing fusion progress so hard.

Who’s got enough power for OpenAI’s 5-gigawatt data center plans?

Five gigawatts is the kind of power a major city needs.

fortune.com

He's asking for over 1% of the entire globe's electricity. Wants the USA to build him enough power plants to run TWO New York states.

Convincing us that it's okay to burn insane amounts of energy cause he'll solve fusion any day now.

And he wants the taxpayers to pay for it.

And they're going to build it because if they don't, the Saudis happily will.

 

[Professor Emeritus]

And they're going to build it because if they don't, the Saudis happily will.

The amount of money they're talking about needing would be nearly 10x higher than Saudi Arabia's entire sovereign wealth fund. And I doubt the Saudis have the infrastructure to build anything like that even if they did have the money.


And if it's that important, then nationalize it. Expending enormous, planet-destroying resources in order to fund profit-obsessed capitalists to create a product of dubious societal value solely because "otherwise someone else might build it first" sounds like a path to collapse.

 

[Professor Emeritus]

He keeps just saying more and more shyt.


1) No we won't

2) "feel" and "many" are so ambiguous that this is a completely meaningless statement

3) Not by any reasonable definition of major



I'm not an expert in those fields to make such predictions, but I seriously doubt Altman knows anything in any of those 3 fields that would justify such claims. He's talking like Musk again.

 

[bnew]

He keeps just saying more and more shyt.


1) No we won't

2) "feel" and "many" are so ambiguous that this is a completely meaningless statement

3) Not by any reasonable definition of major



I'm not an expert in those fields to make such predictions, but I seriously doubt Altman knows anything in any of those 3 fields that would justify such claims. He's talking like Musk again.

The woman cured with gene editing: ‘My supercells have changed my life’

The case of American Victoria Gray shows the hope of new CRISPR therapies, but also their problems: they will cost about three million dollars

english.elpais.com

Teenager's "incurable" leukemia cured by new base editing gene therapy

A teenager in the UK has become the first person to be treated with a revolutionary new gene editing therapy. The patient's incurable leukemia was effectively cleared using "base-edited" immune T-cells from a donor.

newatlas.com

 

[Professor Emeritus]

The woman cured with gene editing: ‘My supercells have changed my life’

The case of American Victoria Gray shows the hope of new CRISPR therapies, but also their problems: they will cost about three million dollars

english.elpais.com

Teenager's "incurable" leukemia cured by new base editing gene therapy

A teenager in the UK has become the first person to be treated with a revolutionary new gene editing therapy. The patient's incurable leukemia was effectively cleared using "base-edited" immune T-cells from a donor.

newatlas.com

I'm well aware of the progression in gene editing - that's why I said, "Not by any reasonable definition of 'major'".

 

[Hood Critic]

The amount of money they're talking about needing would be nearly 10x higher than Saudi Arabia's entire sovereign wealth fund. And I doubt the Saudis have the infrastructure to build anything like that even if they did have the money.


And if it's that important, then nationalize it. Expending enormous, planet-destroying resources in order to fund profit-obsessed capitalists to create a product of dubious societal value solely because "otherwise someone else might build it first" sounds like a path to collapse.

If they were to build the 5-7 data centers he originally proposed, sure.

A single $100 billion dollar data center with the power consumption he's asking for is possible. One of his proposals was to build one in the North Sea and tap into the power of offshore wind turbines - that's a potential 7GW.

No one is going to build a new reactor specifically for OpenAI to utilize, especially while large datacenter operations like AWS and Azure are staking claims to existing unused reactors like those on three mile Island. But sub-sea datacenter research and implementations have been happening in various phases for over a decade now. Cooling accounts for as much as 35-40% of electrical needs in a datacenter.

 

[Professor Emeritus]

If they were to build the 5-7 data centers he originally proposed, sure.

A single $100 billion dollar data center with the power consumption he's asking for is possible. One of his proposals was to build one in the North Sea and tap into the power of offshore wind turbines - that's a potential 7GW.

No one is going to build a new reactor specifically for OpenAI to utilize, especially while large datacenter operations like AWS and Azure are staking claims to existing unused reactors like those on three mile Island. But sub-sea datacenter research and implementations have been happening in various phases for over a decade now. Cooling accounts for as much as 35-40% of electrical needs in a datacenter.

Are we still talking about Saudi doing it? Because even $100 billion (if that's a real number) would be 1/3 of Saudi Arabia's entire annual budget. And if that's a real number, then why were they asking for $7 trillion in funding earlier this year?

And it should be noted again - ALL they are funding at the moment is Altman's hype. What ChatGPT actually does isn't even worth OpenAI's costs, and that's without factoring in the unpaid society costs of enormous energy and resource consumption. The vast majority of what people are actually doing with ChatGPT right now (besides goofing around and cheating at schoolwork) is just predatory pricing to take away real people's work. At the actual cost that it costs, it wouldn't be remotely justifiable unless they're going to do things that have no relation at all to what they currently do.

 

[bnew]

 

[ChatGPT-5]

Racks on racks when this goes public

im on it

 

[Geek Nasty]

AI might legit self destruct On its own. They got away with violating licenses as a non profit but people are coming for checks if they ever go public. Plus AI is already feeding on itself. And Open AI is still hemorrhaging money.

good luck getting the public to pay for a nuclear plant just so you can run AI call centers at a loss.

 

[bnew]

 

[bnew]

Spoiler: thread continued

Spoiler: full text & large images

1/11
@tsarnick
Gary Marcus says he believes OpenAI will become "the most Orwellian company of all time" because they will be forced to turn to surveillance in order to make money



https://video.twimg.com/ext_tw_video/1842320123711119360/pu/vid/avc1/720x720/M4K9tHpHwHSneYiQ.mp4

2/11
@tsarnick
Source (thanks to @curiousgangsta):
https://invidious.poast.org/watch?v=QFMyyAKOOQ0



3/11
@MikeNthDS
only a few months ago he was calling their demise



4/11
@Shawnryan96
AI will be used as surveillance but not because of that. I honestly believe he sits around trying to come with reasons to hate AI. I used to follow Gary back in the day and he was never this bad



5/11
@DanMcAteer88
What happened to deep learning hitting a wall?



6/11
@JWthib
When will he go away?



7/11
@JagersbergKnut
I think ChatGPT already outperforms him at making predictions about the future of AI.



8/11
@opeksoy
wasn’t that their goal, collecting iris data…



9/11
@yaelmendez
They already eat chats and don’t give you a decent search feature.



10/11
@MAD_Logo
Or will it be more like Huxley?

[Quoted tweet]
Orwell v/s Huxley

That last line will haunt me...

11/11
@robdubparker
Yikes




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

 

[Professor Emeritus]

Yet another example of Altman's greed and duplicity:

 

[bnew]

Spoiler: thread continued

Spoiler: full text & large images

1/5
@kimmonismus
OpenAI plans to be profitable ~2029. It can be assumed that OpenAI will be a long-term investment for investors in the VC sector.
In addition, AI is expected to be an excellent deal, especially when agents are established and then billed on a price per use basis similar to SaaS.
In other words, the fact that OpenAI is currently losing $5b should not be overdramatized. As I said before, losses in the venture capital space are completely in line with the norm.

[Quoted tweet]
openai’s cost structure

2/5
@Ed_AgentDev
Losses in VC space are normal, agreed. But $5b is a massive amount to swallow. I'm curious, what's the plan to turn it around by 2029?



3/5
@kimmonismus
On the one hand to drive the prices up to $44 per person and for enterprise up to $1000 for o1 models. That will certainly help.



4/5
@SachGPT23
Well it depends; the market is in state of fragmentation but a shakeout is likely to occur, leading to some perennial oligopoly structure, due to the high FC to compete in this space. If tacit-collusion becomes pervasive as is typical, maintaining a price level (given the cost structure) that is amicable across other players, you have a business with sustainable profits.



5/5
@NoesisML
The real takeaway here is that OpenAI’s losses aren’t driven by inference costs




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196

 

[bnew]

Spoiler: full text & large images

1/1
@danwolch
ChatGPT has 6x the search traffic of Gemini and almost 20x Claude.

They're capturing more LLM chat data than anyone else. In a world of data network effects, acquiring memory about you, and the cost of inference going to 0, their lead would only widen, wouldn't it?

To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196