Original video in Chinese.
Key Takeaway
- Elon Musk has sued Sam Altman and OpenAI, accusing them of abandoning OpenAI’s founding values of “benefiting all humanity” and “open source,” and shifting toward maximizing profits.
- The core of the lawsuit is OpenAI’s agreement with Microsoft and whether OpenAI has already achieved AGI and privatized it.
- Elon Musk believes OpenAI’s AGI technology, such as GPT-4 and the secretly developed model Q, has been controlled by Sam Altman and funneled to Microsoft, betraying its non-profit origins.
- This lawsuit is not only about OpenAI’s future, but may also affect the direction and ethical norms of the entire AI industry.
Elon Musk suing Sam Altman is absolutely this year’s biggest drama.
In the complaint filed with the California Supreme Court, he is the sole plaintiff, suing Sam and OpenAI all at once. And he has also requested a jury trial.
In other words, Elon Musk wants to drag Sam and OpenAI out of the shadows and into the light — in court, under oath, they’ll have to explain everything clearly, and leave a record. This includes:
- What exactly happened with the attempted power grab and rebellion a few months ago?
- What exactly is going on with Microsoft?
- Has the mysterious project Q actually achieved AGI?
This trial will affect not only Sam personally, and the two companies OpenAI and Microsoft, but also the entire Silicon Valley ecosystem and the direction of AI.
This complaint is a bit long, 46 pages. I’ll go through it with you first. Elon Musk’s logic is very simple. He answers two questions in plain language:
- Why
- Why Now
Why: Betraying Values
Many people think profit comes before everything else. But in elite circles, values come first, and they are non-negotiable.
OpenAI is the product of a set of values.
Elon Musk has always firmly believed that Superman Machine Intelligence, super machine intelligence, is the greatest threat to all of humanity.
As deep learning made substantive progress and became increasingly practical, that threat drew closer and closer.
The trigger was Google’s acquisition of DeepMind in 2014. That acquisition put Google far ahead on the road to AGI.
Elon Musk believed that if artificial general intelligence technology were controlled by a company like Google, driven by greed and profit, then our entire human society would be finished.
So in 2015, he, Altman, and Brockman jointly decided to establish a non-profit research organization:
- The purpose of developing artificial general intelligence was to benefit all humanity, not maximize profits — that was the mission.
- Unless security concerns were involved, research results had to be open source — that’s where Elon Musk came up with the name OpenAI.
OpenAI’s birth had a very Avengers-like feel: for all humanity, against the evil Google.
Elon Musk, like Iron Man, provided tremendous help in OpenAI’s early stage, especially by bringing in top global talent, such as the chief scientist.
In the beginning, OpenAI really was open. But according to the arc of a movie plot, this was the moment when the reversal should have come.
Pay attention to these dates, which the complaint specifically emphasizes:
- In 2019, Sam became CEO of OpenAI.
- Elon Musk’s support for OpenAI continued until September 14, 2020.
- A few days later, on September 22, 2020, Microsoft came in.
At that time, OpenAI provided Microsoft with the GPT-3 model. At the same time, OpenAI also made the training method for GPT-3 public, allowing others to build similar models.
There is one very important clause in OpenAI’s agreement with Microsoft:
Microsoft can only obtain Pre-AGI technology.
In other words, one day when OpenAI truly achieves AGI, that technology will not be given to Microsoft.
So who decides whether newly developed technology counts as AGI?
The board.
If the board defects and defines AGI technology as non-AGI technology, then it can continue to provide technology to Microsoft in a justified way without violating the agreement, in exchange for financial returns.
That is why Elon Musk insists on doing this now.
There is a traitor in the revolutionary ranks, colluding with the enemy to steal the fruits of the revolution!
Why Now: AGI Has Been Achieved
In March 2023, OpenAI released GPT-4. I don’t need to say much about how powerful it is. Even after a full year, no company has yet produced a large model at the same level.
But OpenAI has never disclosed the design details of GPT-4. While we know nothing about GPT-4, Microsoft has reaped enormous benefits from it, with GPT-4 embedded in its products.
More importantly, Microsoft’s researchers said that GPT-4 is already an early version of an artificial general intelligence system.
So starting with GPT-4, OpenAI under Sam’s leadership has drifted further and further off course:
- Completely not open source.
- Continually providing Microsoft with exclusive technology.
What really made Elon Musk unable to sit still was this:
According to information leaked from inside, OpenAI’s secretly developed model Q has seen a major performance jump and is very likely to have already achieved AGI.
However, last November, the board originally wanted to fire Sam, but he turned the tables and wiped out the entire board in one fell swoop. The current board members were all personally chosen by Sam, and they are close to Microsoft.
This means Sam has cleared away all obstacles. The research results that were supposed to be open source can now be delivered to commercial entities like Microsoft.
And didn’t Sam also want to raise 7 trillion? Then he can likewise use technology as bargaining chips to secure backing from even bigger sources of capital than Microsoft, such as in the Middle East.
Going one step further, if Sam and Microsoft’s approach works, then in the future everyone will choose to start companies in the form of non-profit research organizations. VC money can go in, while on one hand enjoying policy subsidies like tax breaks, and on the other hand exclusively enjoying the research results. That harm is also enormous.
So this lawsuit involves technology, business, legislation, and also drama like betrayal, power grabs, and value transfers. It is bound to be a global hot topic. With such high attention and the judicial system stepping in, OpenAI will no longer be able to hide things — and that is precisely their weakness.
Elon Musk really is a top-tier master manipulator.