In June 2015, during his 44th birthday, Elon Musk hosted a lavish three-day party at a Napa Valley villa, attended by friends and family. In the middle of the celebration, they say that around a bonfire, an intense debate arose that still resonates in the present.
The discussion was between Musk and Larry Page, co-founder of Google. Musk defended the need to implement safeguards in the evolution of AI. Instead, Page believed that the next step in evolution was for machines and humans to merge allowing for post-human life, in the cloud.
The discussion became even more intense when Page accused Musk of being a “speciesist,” that is, privileging humanity over other possible forms of intelligence. Musk, for his part, responded by stating that yes, he was “pro-human.”
This confrontation not only marked a breaking point in their friendship, but also led Musk to found OpenAI that same year, to develop AI “safe and accessible to everyone”, although we do not believe in Musk’s benefits beyond power and economic gain. This event became a crucial milestone in the history of the development of artificial intelligence.
Musk in search of “safe” AI for humans
After the controversy, Elon Musk contacted Sam Altman, known for his ability to identify and support innovative startups and for his vision of the technological future.
Musk and Altman, along with other prominent entrepreneurs and scientists such as Greg Brockman, Ilya Sutskever and Wojciech Zaremba, decided to found OpenAI in December 2015.
The idea of OpenAI was born from a shared concern for the safe and beneficial development of artificial intelligence. Their goal was to ensure that AI was developed in a way that would benefit all of humanity, rather than just a few.
To achieve this, OpenAI was established as a non-profit organization, with the purpose of moving forward in a way that maximizes the benefits for humanity as a whole, although this, as we will see, ended by creating a for-profit subsidiary.
Google at the forefront of AI
Years before OpenAI was founded, Google had already taken significant steps to lead and stay at the forefront of AI. In 2011, Google launched the Google Brain project, a part-time collaboration between Jeff Dean, Greg Corrado, and Stanford professor Andrew Ng.
This project focused on developing techniques to solve complex AI problems. Google Brain leveraged Google’s vast cloud computing infrastructure to build large-scale deep learning systems.
In January 2014, Google acquired DeepMind, a British AI startup, for more than $500 million. Founded in 2010 by Demis Hassabis, Shane Legg, and Mustafa Suleyman, DeepMind was dedicated to developing general learning algorithms applicable to a variety of problems.
As a subsidiary of Google, DeepMind was notable for creating AlphaGo, the AI program that beat the world champion of the game Go in 2016.
In 2015, Google launched TensorFlow, an open source library for machine learning that quickly became an essential tool for AI researchers and developers around the world.
Google has also integrated AI into many of its products, such as Google Photos, which uses image recognition, and Google Assistant (Ok, Google), which uses natural language processing.
These initiatives helped Google consolidate its leadership and compete with other startups like OpenAI.
Elon Musk changes his mind and breaks with OpenAI
Initially, Elon Musk supported the idea of OpenAI being a non-profit organization. However, they soon realized that the resources required to develop artificial general intelligence (AGI) were much greater than they had anticipated.
Musk and other members of OpenAI acknowledged that they would need billions of dollars to compete with giants like Google and its DeepMind. In 2017, they began considering turning OpenAI into a for-profit entity to secure the necessary resources.
One of Musk’s proposals was to merge OpenAI with Tesla or for he to have full control of the organization. However, this idea was met with resistance from the other co-founders, such as Sam Altman and Greg Brockman, who felt it went against OpenAI’s original mission.
Additionally, there were concerns about potential conflicts of interest, given that Musk was also developing AI technologies at Tesla. In February 2018, Musk decided to leave OpenAI (and take all his money) with the intention of creating a competitor to Google through Tesla.
OpenAI runs out of money and Microsoft throws a lifeline
After Elon Musk’s departure, OpenAI faced several financial challenges. The organization realized that it would need an immense amount of resources to compete in the development of advanced AI.
In 2019, OpenAI made the strategic decision to become a limited for-profit entity, known as OpenAI LP. This hybrid structure allowed them to attract significant investments while maintaining their original mission.
In July 2019, Microsoft invested $1 billion in OpenAI, providing the necessary resources to continue its research and development. It also allowed OpenAI to access Azure supercomputing infrastructure, which was essential for training its advanced AI models.
Meanwhile… on Facebook
In 2013, Mark Zuckerberg, the founder and CEO of Facebook, along with Mike Schroepfer, CTO, and other company leaders, recognized the transformative potential of artificial intelligence.
Thus, in September 2013, Facebook announced the creation of Facebook AI Research (FAIR), with laboratories in Menlo Park, New York and London. To lead FAIR, they hired Yann LeCun, a pioneer in the field of deep learning and convolutional neural networks.
LeCun, known for his work in image recognition, took on the role of director and began building a world-class team of researchers. In 2016, FAIR launched PyTorch, an open source deep learning library that quickly became an essential tool for AI researchers and developers.
They also developed tools such as fastText for natural language processing and Detectron for object detection, both open source.
The failed bet on the metaverse
In 2021, Facebook changed its name to Meta to reflect its new vision focused on the “metaverse.” Mark Zuckerberg presented it as the “next great frontier of the digital world”: a shared virtual environment accessible through virtual reality (VR) and augmented reality (AR) devices.
Meta invested billions of dollars in this vision, including the creation of Reality Labs, the team charged with developing VR and AR technologies. However, despite large investments, the metaverse did not immediately achieve the expected impact.
In 2022, Meta reported significant losses at Reality Labs, with more than $9.4 billion in the first three quarters of the year. These losses, along with the lack of adoption of the metaverse, sparked criticism over whether resources were being diverted from important areas, such as AI research.
OpenAI takes the lead
The OpenAI team, backed financially and with technological support from Microsoft, moved forward at a dizzying pace. In June 2020, OpenAI launched GPT-3, an advanced language model that surprised with its ability to generate coherent text from simple prompts.
GPT-3 quickly became an essential tool for developers and businesses. Its API was integrated into Microsoft’s Azure platform, allowing users to access its capabilities through the cloud.
In 2021, OpenAI introduced two significant innovations: DALL·E, a model capable of generating images from textual descriptions, and CLIP, a model that can effectively understand and relate images and text. These advances demonstrated the potential of artificial intelligence to create visual content and improve multimodal understanding.
Also in collaboration with Microsoft, OpenAI launched GitHub Copilot in 2021, a tool that uses GPT-3 to assist developers in writing code faster and more accurately. This integration showed how AI could transform productivity in software development.
The big milestone came in November 2022 with the launch of ChatGPT, a chatbot based on the GPT-3.5 model. This launch marked a before and after in the field of AI. The world was amazed to see how ChatGPT could maintain fluid conversations, answer complex questions, and generate creative content. Nothing would be the same again.
Crisis at OpenAI, Altman is fired
In November 2023, OpenAI was hit by an unexpected crisis. On November 17, the company’s board of directors announced the dismissal of its CEO, Sam Altman. The news surprised many, as Altman had been a key figure in the growth and success of OpenAI.
The council, which included figures such as Ilya Sutskever, Adam D’Angelo, Tasha McCauley and Helen Toner, justified its decision by arguing that Altman had not been “consistently sincere” in his communications with them.
They alleged that this lack of transparency complicated their ability to fulfill their responsibilities and maintain confidence in their leadership. Additionally, concerns were raised about the handling of safety in AI and allegations of abusive behavior were raised.
Employees and Microsoft to the rescue
Altman’s dismissal generated a rapid and forceful reaction. More than 500 of OpenAI’s 700 employees signed a letter demanding the board’s resignation and Altman’s reinstatement.
Pressure also came from outside the company, with Microsoft, OpenAI’s main investor, stepping in to support Altman. Microsoft CEO Satya Nadella expressed concern and supported Altman’s return.
After five days of intense negotiations and leadership changes, on November 22, 2023, Sam Altman was reinstated as CEO of OpenAI.
The original board was replaced by a new board of directors, led by Bret Taylor, former co-CEO of Salesforce, and consisting of Larry Summers, former US Secretary of the Treasury, and Adam D’Angelo, the only member who remained from the previous board.
Ilya Sutskever, who had notified Altman of his dismissal, left his position on the board, but continued to work at the company.
Open vs. open models closed models
The evolution of language models has been one of the most fascinating and controversial areas in the field of artificial intelligence. Two different approaches have emerged here: open language models, such as Meta’s LLaMA, and closed models, such as OpenAI’s GPT.
GPT-3, developed by OpenAI, was integrated into Microsoft’s Azure platform, allowing users to access its capabilities through the cloud. However, OpenAI chose to keep the model closed, limiting access through a controlled API.
This decision sought to protect the technology and ensure responsible use, but also restricted accessibility.
In contrast, Meta took a different approach. In 2023, Meta released LLaMA, a pre-trained large language model, followed by LLaMA 2, which is freely available for commercial and research use.
Mark Zuckerberg, CEO of Meta, has been a strong advocate of open source models. In a letter published in July 2024, Zuckerberg shared his vision on the importance of open models for the future of AI.
Zuckerberg compared the development of AI to the evolution of operating systems. He recalled how, in the early days of computing, major technology companies developed closed versions of Unix.
However, over time, Linux, an open source operating system, gained popularity due to its flexibility, security, and the extensive ecosystem that supported it. Today, Linux is the industry standard for cloud computing and mobile operating systems.
Zuckerberg argued that AI will follow a similar path. According to him, open source models will allow more people around the world to access the benefits and opportunities of AI. This approach will prevent power from being concentrated in the hands of a few companies and will allow technology to be deployed more equitably and safely in society. However, like the rest of the protagonists in this story, they tell us what we want to hear, the beautiful part in which they look good. Most likely, Zuckerberg admitted that the first versions were open source to put himself at the top of the AI race, which he had lost due to his failed bet on the metaverse.
There is always time for one more lawsuit
The relationship between Elon Musk and OpenAI has been marked by ups and downs, with disagreements and breakups that have recently escalated into a legal sphere. In March 2024, Musk filed a lawsuit against OpenAI and its former CEO, Sam Altman.
In his lawsuit, Musk alleged that Altman and other co-founders had abandoned the organization’s founding principles, prioritizing profits over the public good.
Musk claimed that he was tricked into co-founding OpenAI under the promise that the entity would always be nonprofit, focused on the security and openness of artificial intelligence. However, OpenAI responded by publishing emails from Musk that demonstrated his previous support for the idea of converting the organization into a for-profit entity to secure the necessary resources.
These emails contradicted Musk’s allegations and showed that he had approved the structural transformation of OpenAI.
In June 2024, Musk unexpectedly withdrew his lawsuit, without providing a clear explanation. However, the dispute continued and in August 2024, Musk revived his lawsuit, again accusing OpenAI and Altman of fraud and diverting the organization from its founding goals.
In this new phase of the lawsuit, Musk was more explicit in his accusations. He claimed that Altman and his “accomplices” had used altruism as a lure to attract their investment, only to change the organization’s course as technology advanced and profits became more attractive.
A conclusion… for now
The recent history of AI illustrates how agreement and disagreement between brilliant minds has shaped the advancement of technology.
From the founding of OpenAI and its collaboration with Microsoft to the legal controversies involving Elon Musk, both cooperation and conflict have been present at each stage.
On the one hand, the alliance between OpenAI and Microsoft has demonstrated how strategic partnerships can accelerate technological progress. At the same time, disagreements, such as Musk’s lawsuits, have underscored the ethical and financial complexities involved in developing AI.
Technological advancement does not follow a linear path; It is full of debates, challenges and difficult decisions. Tensions between OpenAI, Microsoft, Google, Meta and other entities have shown that disagreement can be as productive as consensus.
As we explore the frontiers of technology, it is critical to recognize that both collaboration and conflict are essential drivers of progress. To know more about the history of AI, consult this article where we tell you about the beginnings, also very interesting, about the 1950s
This post is also available in: