OpenAI Reveals Elon Musk’s Emails

OpenAI made very harsh statements about Elon Musk, who took it to court. While explaining how the relationship between the parties developed, the e-mails sent were also shared. A response also came from Elon Musk.

In a news we shared with you a few days ago, we talked about a tension between OpenAI and billionaire businessman Elon Musk. we mentioned. Elon Musk complained about OpenAI and the company’s CEO, Sam Altman, He was on his way to the court. Musk and his lawyers said that both OpenAI and Sam Altman had moved away from their goals, and that this was due to their agreement. it is against he was saying.

Now, there has been a remarkable development regarding the issue. OpenAI management broke its silence and criticized Elon Musk’s claims. rejecting made a comprehensive statement. This statement touched upon issues such as what OpenAI has done so far and why it started to attach importance to materiality. Also, the conversation between Elon Musk and OpenAI emails was also shared. Elon Musk also responded to this. Let’s examine this issue in depth together.

Things changed when OpenAI realized how costly artificial general intelligence was

OpenAI says its first brush with Elon Musk occurred at the end of 2015. The founders of the company, Greg Brockman and Sam Altman, were $100 million He was planning to raise funds. However, this figure was seen as quite low for Elon Musk. According to Musk, the team should have started with a budget of $1 billion. At least he should have said so. Meanwhile; According to OpenAI’s statement, Elon Musk joined the company at the first stage. $45 million He gave around .

In the first stage, non-profit OpenAI in 2017 he notices. So much so that it is not possible to develop general artificial intelligence with current budgets. Realizing how much computing power they need, the team understands that they have to spend billions of dollars for this. OpenAI said in a statement that this was Elon Musk’s even more than you expected emphasizes that it is.

Elon Musk has known since those days that OpenAI must be for-profit.

Elon Musk Sam Altman

Realizing something, the team met with Elon Musk in 2017 and for profit They say they want to become a new formation. Elon Musk says they can do this, but gives OpenAI two options. These are either Merger with Tesla or putting Elon Musk in charge of the formation. However, OpenAI cannot reach an agreement with Elon Musk, thinking that the formation cannot be established under a single leader. In the statement, Elon Musk’s financial support was also provided while all this was happening. cut It is stated.

“We are expanding our mission with artificial intelligence tools that are accessible to everyone.”

Although OpenAI now admits that they are for profit, they want people to be aware of their work. In this context; on the one hand open source The team states that they support projects and that the tools they offer free of charge have users from all over the world. Giving some examples, OpenAI states that Albania uses artificial intelligence to become compatible with the EU. A company called Digital Green provides agricultural extension services with OpenAI services. 100 times cheaperIt is stated that this way, the income of farmers in Kenya and India is increased.

A.I.

OpenAI, Elon Musk’s latest attempt at artificial general intelligence It is not just open source also says. Ilya Sutskever even wrote to Elon Musk in an email:As we get closer to developing AI, it makes sense to start being less open.” he says. Elon Musk responded to this by saying “Yes” he uses the expression.

Let’s talk about the e-mails between the parties:

Sam Altman

Part of Elon Musk’s first email to Greg Brockman (November 22, 2015)

We need to go with a much larger number than $100 million so we don’t look hopeless compared to what Google or Facebook are spending. I guess we should say that we started with a $1 billion funding commitment. This is real. I will cover anything that no one else does.

Email sent to Elon Musk explaining why OpenAI is for-profit (January 31, 2018 – email sender withheld)

Working at the cutting edge of AI is unfortunately expensive. For example, Google, in addition to DeepMind, also has Google Brain, Research and Cloud. And they own TensorFlow, TPUs, and about a third of all research (they even hold their own AI conference).

Moreover, achieving artificial general intelligence is not easy. If historical trends are any indication, progress in artificial intelligence is primarily driven by systems such as computing, data, and infrastructure. The basic algorithms we use today have remained largely unchanged since the 90s. Not only that, any algorithmic advance published in a paper somewhere can be reimplemented and incorporated into a project almost instantly. Conversely, algorithmic advances alone are ineffective without scale, making them scary.

It seems to me that OpenAI is just wasting money today and its funding model cannot reach the scale to seriously compete with Google. If you can’t compete seriously but keep working, you may actually be making things worse and helping them “for free.” Because it is very easy for them to copy any progress and immediately incorporate it on a large scale.

A for-profit entity can create a more sustainable revenue stream over time and possibly bring in a lot of investment with the existing team. But building a product from scratch would steal the focus of AI research, take a long time, and it’s unclear whether a company can “catch up” to Google’s scale. The most promising option I can think of would be OpenAI connecting to Tesla, as I mentioned before. I believe tie-ups with other large companies (like Apple and Amazon) will fail due to incompatible company DNA.

Let me make a “rocket” analogy to this. Tesla had already built the “first stage” of the rocket, with the Model 3’s entire supply chain, onboard computer, and a permanent internet connection. The “Second stage”, which will enable the rocket to be launched, will be a fully autonomous solution based on OpenAI’s large-scale neural network training. We could sell a lot of cars/trucks with a functioning fully autonomous driving solution in about 2-3 years. If we do this really well, we will increase Tesla’s market value and we can use that revenue to fund AI work at scale.

I don’t see any other path that has the potential to achieve sustainable capital at Google’s scale within a decade.

Elon Musk responds to this e-mail as follows (February 1, 2018)

Elon Musk

Tesla is the only way OpenAI can challenge Google. However, even in this case, the possibility of being a counterbalance against Google is not zero, but very low.

Elon Musk sends another email to Ilya Sutskever (December 26, 2018)

Without a dramatic change in implementation and resources, my assessment of the likelihood of OpenAI competing with DeepMind/Google is 0%. Not even 1%. I wish it were otherwise. Even raising a few hundred million won’t be enough. Either billions of dollars must be spent annually or forget about it.

To the statement made by OpenAI and all the e-mails shared from here accessible.

Elon Musk’s answer: “Change your name”


source site-34