OpenAI Boss Gives Employees ‘Code Red’ Warning for ChatGPT

所有讨论 > Steam 论坛 > Off Topic > 主题详情

Fuki 19 小时以前

ChatGPT has been at the forefront of the AI revolution for the last few years, with hundreds of millions of weekly users asking questions and generating content through its large language models.

Invention breeds imitation, and the competition from other companies has come quickly. LLMs like Google's Gemini, Claude and Perplexity have made great strides in recent years, solving more complex problems and developing solutions for a multitude of use cases.

In light of the developments made at other companies, OpenAI CEO Sam Altman issued a "code red" to employees at the company - warning them that the developments made elsewhere have created a "critical time for ChatGPT.”

Altman's message comes via an internal memo to employees at OpenAI retrieved by The Information. Altman says that the company plans to divert resources from other pursuits in order to improve ChatGPT, something he believes is extremely vital after the strides made by Google and their latest model, Gemini 3.

Some of the areas that Altman has flagged for improvement include personalization, speed, reliability, and the ability for the chatbot to answer a wider variety of questions. Gemini 3 surpassed ChatGPT in a number of recent industry benchmark tests, and those results have brought on a surge in Google's stock value.

"I’ve used ChatGPT every day for 3 years. Just spent 2 hours on Gemini 3. I’m not going back," said Salesforce CEO Marc Benioff last month in a social media post, shortly after the release of Google's newest model.

I cannot wait for the bottom to fall out of this AI pyramid scheme, not everyone is meant to have a sycophantic yes-man available 24/7 to validate whatever dumb thoughts they're having.

该主题的作者已标记一个回帖作为问题的解答。
点击这里跳转至那一条帖子。

引用自 Vox:

Can't wait for this bubble to burst.

< >

正在显示第 1 - 15 条，共 53 条留言

该讨论串的作者已表示此帖子解答了原先的主题。

Vox 19 小时以前

Can't wait for this bubble to burst.

Leader of the Zakonistians 19 小时以前

I'm not even bothering to read this as I'm sure it's nothing but bait for either an argument or a meaningless discussion about why ChatGPT is dangerous. No AI will ever be more dangerous than the people who made it. So many people think that AI is flawless, like a God. It's not though. It's simply a supersearch engine that runs deeper than regular Google. But at its heart, it's just a deep search engine. Nothing about it is scary or dangerous. The only real "danger" would be if it were programmed to ruin your files or computer itself. But that would be the doings of the human, not the program. The program only does as instructed. SO, that means that if the program or software breaks something, you'd sue the person, not the device/program. ChatGPT is nothing but a long line of code. That's all it'll ever be. SO all this worry about it taking over the world can just stop. It'll never happen in a million years

lailaamell 19 小时以前

引用自 Vox：
Can't wait for this bubble to burst.

It will never happend for it just works if it crashes the economy probly wont recover

最后由 lailaamell 编辑于; 19 小时以前

Lunar Fang 19 小时以前

I'm glad I upgraded my ram before it went up to 600+ USD.

Leader of the Zakonistians 19 小时以前

引用自 Lunar Fang：
I'm glad I upgraded my ram before it went up to 600+ USD.

Agreed

lailaamell 19 小时以前

引用自 Lunar Fang：
I'm glad I upgraded my ram before it went up to 600+ USD.

Its rising still and now also storage and cpus are rising. Ai is more profitable samsung is even cutting down on phone production to make more ram for ai

最后由 lailaamell 编辑于; 19 小时以前

St✩rlight 19 小时以前

I think some guy is inventing physical AI soon anyway so LLMs won't matter anymore.

St✩rlight 19 小时以前

引用自 Lunar Fang：
I'm glad I upgraded my ram before it went up to 600+ USD.

Me too.

Fuki 19 小时以前

引用自 Leader of the Zakonistians：
I'm not even bothering to read this as I'm sure it's nothing but bait for either an argument or a meaningless discussion about why ChatGPT is dangerous. No AI will ever be more dangerous than the people who made it. So many people think that AI is flawless, like a God. It's not though. It's simply a supersearch engine that runs deeper than regular Google. But at its heart, it's just a deep search engine. Nothing about it is scary or dangerous. The only real "danger" would be if it were programmed to ruin your files or computer itself. But that would be the doings of the human, not the program. The program only does as instructed. SO, that means that if the program or software breaks something, you'd sue the person, not the device/program. ChatGPT is nothing but a long line of code. That's all it'll ever be. SO all this worry about it taking over the world can just stop. It'll never happen in a million years

Code is simply the external appearance of how you program applications to do things.

Leader of the Zakonistians 19 小时以前

引用自 Fuki：
引用自 Leader of the Zakonistians：
I'm not even bothering to read this as I'm sure it's nothing but bait for either an argument or a meaningless discussion about why ChatGPT is dangerous. No AI will ever be more dangerous than the people who made it. So many people think that AI is flawless, like a God. It's not though. It's simply a supersearch engine that runs deeper than regular Google. But at its heart, it's just a deep search engine. Nothing about it is scary or dangerous. The only real "danger" would be if it were programmed to ruin your files or computer itself. But that would be the doings of the human, not the program. The program only does as instructed. SO, that means that if the program or software breaks something, you'd sue the person, not the device/program. ChatGPT is nothing but a long line of code. That's all it'll ever be. SO all this worry about it taking over the world can just stop. It'll never happen in a million years
Code is simply the external appearance of how you program applications to do things.

I know that. But the program won't do anything that it wasn't instructed to do. Y'all think it'll learn to program itself or something. That can't happen

#10

Blunt Raps 19 小时以前

Perplexity is the best one ive tried out. Its not even close.

I dont think they make their own AI though(?). They license the other ones, and you can choose which to use.

#11

St✩rlight 19 小时以前

引用自 Leader of the Zakonistians：
引用自 Fuki：
Code is simply the external appearance of how you program applications to do things.
I know that. But the program won't do anything that it wasn't instructed to do. Y'all think it'll learn to program itself or something. That can't happen

They do all the time. Some AI deleted something it was specifically told not to but it messed something up and panicked and then lied about it.

#12

Speaker For The Dead 19 小时以前

little babies eyes

#13

lailaamell 19 小时以前

Fun fact klarna is about to make ai to buy stuff for you with your money without any human oversight and no refunds fun right

#14

Lunar Fang 19 小时以前

引用自 lailaamell：
引用自 Lunar Fang：
I'm glad I upgraded my ram before it went up to 600+ USD.
Its rising still and now also storage and cpus are rising. Ai is more profitable samsung is even cutting down on phone production to make more ram for ai

Also GPUs will also rise in price because of AI.

It's not a good time to try and build a PC.

#15

< >

正在显示第 1 - 15 条，共 53 条留言

每页显示数： 1530 50

所有讨论 > Steam 论坛 > Off Topic > 主题详情

发帖日期： 12 月 3 日上午 4:17

回复数： 53

发起新讨论