The changes here are quiet

Musk's so-called channels refer to connections with international open-source organizations.

As the number one madman in the tech world, one of the main drivers behind the commercialization of mobile payments and electric vehicles, and a pioneer in private spaceflight, Musk already has myriad ties to the open-source community.

When the research and development resources required for advanced technology become increasingly astronomical, open source becomes the only way for everyone to contribute and achieve great things.

When Musk founded OpenAI, he originally intended to embrace the open-source route, but later OpenAI turned into CloseAI, and his current attempts to push OpenAI back to open source have also failed.

The optimal window for OpenAI to return to open source has already passed.

And this window was personally closed by the Wall Street crowd, who also came from Western civilization.

To make a hefty profit from shorting, they gave DS a push in the right direction. If Musk were to handle this situation, he should have ignored DS immediately after realizing its significant benefits, then quickly used DS to pressure OpenAI's investors to open-source OpenAI.

As long as OpenAI became open source in time, DS would still be irrelevant.

Because DS itself is open source, OpenAI could exploit a time difference, leverage the discourse advantage controlled by the West, and desperately squeeze out a time window to copy DS.

Then, in turn, accuse DS of copying OpenAI.

This way, DS's advantage of low training costs would be gone.

As for who copied whom, how could it be clear in a short while?

Figuring out who copied whom isn't actually important; whether OpenAI can return to the open-source track is also just a superficial matter. What's important is the development path of AI.

In Musk's view, Dragon Country has already taken the path of embracing open source and also has the strength to take this path. Therefore, their only strategy is to steadfastly continue on the open-source path, relying on their first-mover advantage in the open-source community.

So, the situation is actually very strange.

Everyone knows that AI development relies on open source, or rather, the IT industry itself has benefited from open source to achieve its great development today.

This is equivalent to Eagle Nation abandoning its former path to success. Then Dragon Country naturally took the path that Eagle Nation once succeeded on.

A-Long is also very confused.

"I don't know either, why did Eagle Boy give up on himself? I still wanted to compete with it in the open-source world to see who was more of a white lotus. But, well, it gave up first."

"I also don't know why Eagle Boy started by stabbing itself twice."

.

.

Looking globally, there are only two entities with the strength to invest in AI. If one of them chooses to play closed source, then everyone in the world can only follow the other one in playing open source. Because people have no choice; in the face of real money, all ideologies and values can go to hell.

Even more terrifying is that open source will also bring about an increase in soft power.

Behind open source is "knowledge without defense"; humanity's yearning for fairness is deeply buried in their hearts, and knowledge equality is the starting point for all fairness.

Therefore, open source will bring about a great deal of value alignment.

Since open source has so many benefits, why did OpenAI move towards closed source?

Because the benefits of open source are long-term and apply to the entire collective. But the benefits of closed source are immediate and directly visible to OpenAI's investors.

It is precisely because of this that the bloodthirsty jackals of capital in Eagle Nation would not hesitate to brutally stab their own country's AI industry twice, because stabbing is not the goal; the short-selling profits brought by the stabbing are the goal.

Even if this move pushes DS directly to the throne of international AI open source. But who cares, the dollars in the pocket are the most real. Anyway, doing this is all legal in Eagle Nation.

Ultimately, Musk's attempt to return OpenAI to open source failed.

To compensate, as Lin Yuan expected, Musk started telling stories again.

Thus, Grok-3 was hastily pushed out as a firefighter.

Although Musk officially resigned from OpenAI's board in 2018, he still has extensive collaborations with OpenAI, both as a key co-founder and for the integration of the AI field.

But now, Musk can officially abandon OpenAI.

As a backup, Grok-1 was partially open-sourced after its launch in 2023.

Grok-1 uses the Apache 2.0 license. It allows free commercial use but requires attribution and prohibits military use. Additionally, it specifically requires that any models adjusted based on Grok-1 must also be open source.

It can be said that Grok-1 intended to be open source.

However, by Grok-1.5, the original intention of open source had to be paused again.

The reason is still the old problem—investors are eager for returns.

Given the current development efficiency of the West and their brute-force AI training technical route, Musk cannot fund an entire AI large language model on his own.

This is actually not just about money; it also involves the issue of sharing benefits. Not allowing others to invest means not being willing to share the pie with others.

The only thing Musk can do is try his best to control Grok's leadership, avoid repeating OpenAI's mistakes, and also keep Grok as a backup and a whip constantly pushing against OpenAI's ChatGPT.

Because investors were eager to see returns, Grok was also required to launch a paid API as soon as possible. After the model deployment is complete, a paid API is a money printer; which investor can resist such temptation?

Of course, this basically means being incompatible with open source.

Therefore, Musk had not particularly pushed Grok before. Because earning money through paid APIs is truly not what a promising AI large language model should look like.

In a word, doing so looks too unseemly.

But after ChatGPT was severely slapped by DS, to keep ChatGPT alive, its investors ultimately had to agree to embrace open source in the next version.

Similarly, Musk's Grok also lost the resistance from those who dissed open source.

So at this critical juncture, Grok-3 entered the stage as a firefighter. And Musk also guaranteed that Grok-3 would be open source a few months after its release.

Lin Yuan also watched the Grok-3 launch event. Awkward, capital A awkward!

Because the technical focus of the press conference was on model scoring.

Scoring, after all, is a very mystical thing. It's even possible to optimize some models in advance for the scoring rules.

Thus, Grok-3 naturally led DS-R1 in various scores.

Then a group of highly intelligent guys blabbered for a long time in front of the camera, just praising Grok's prowess, but deliberately ignoring an important factor—talking about output without discussing input is acting like a hooligan.

Grok-3 only dared to talk about scores, but not about output.

Lin Yuan muttered to himself, 'As for scores, with enough effort and without considering training costs, it's always possible to create a version of the model specifically for scoring and evaluation.'

This is like car manufacturers submitting vehicles for inspection; if the manufacturer is willing, they can entirely create a few cars specifically for testing.

As for how much the car would sell for in the market based on the standards of the submitted inspection vehicles, they just play dumb and pretend not to know.

Grok-3's release, in Lin Yuan's opinion, is a very important sign. Because it signifies that the Eagle Nation's ship can no longer turn around. Because they no longer have the courage to acknowledge others' excellence.

Because only when one lacks confidence in surpassing others will they dare not acknowledge others' excellence.