Lifeboat Foundation

The GPT phenomenon and the future of humanity in the face of advances in Artificial Intelligence.

The Age of Artificial Intelligence is an increasingly present reality in our daily lives. With the rise of technologies such as Natural Language Processing (NLP) and Artificial Neural Networks (ANN), the possibility of creating machines capable of performing tasks that were previously exclusive to humans has emerged.

One of these technologies is the Generative Pre-trained Transformer, better known as GPT. It’s the Large Language Model (LLM) developed by OpenAI.

OpenAI was founded in San Francisco, California in 2015 by Sam Altman, Reid Hoffman, Jessica Livingston, Elon Musk, Ilya Sutskever, Peter Thiel, among others, who collectively pledged $1 billion. Musk resigned from the board in 2018, but continued to be a donor to the project.

In 2019, when OpenAI became a for-profit organization, Microsoft invested $1 billion, followed by a second multi-year investment in January 2023, reported to be $10 billion.

Microsoft Bing versus Google Search…

The original non-profit organization aimed to develop Artificial Intelligence in a safe and beneficial way for humanity. The founders were concerned that Artificial Intelligence could represent an existential threat to humanity if not developed responsibly.

Thus, the initial mission of OpenAI was to conduct research and develop Artificial Intelligence technologies that would benefit humanity as a whole, rather than just profiting from this technology.

One of the main reasons for the change was that OpenAI realized it would need much more financial resources to achieve its ambitious goals of developing advanced Artificial Intelligence that can be used safely and beneficially. Becoming a for-profit organization would allow OpenAI to attract significant investments and establish commercial partnerships with other companies, as has been seen with Microsoft.

By the way, to give an idea of the costs involved, OpenAI’s systems run on the fifth most powerful supercomputer in the world. ChatGPT, for example, was trained on NVIDIA GPUs. It is rumored that 10,000 NVIDIA GPUs were used to train the ChatGPT model.

Currently, ChatGPT alone requires 30,000 NVIDIA A100 GPUs to operate. ChatGPT uses a different type of GPU than the ones people put in their gaming PCs. Just one NVIDIA A100 GPU costs between $10,000 and $15,000.

The hardware is keeping up well with the evolution of software. Nevertheless, the OpenAI system is now experiencing disruptions after an explosive growth in users and the numerous requests that simultaneously inquire the LLM, suggesting that this clearly is not enough capacity. As someone once said, innovation has a cost!

The creation of monsters that could be angels or demons… or even both, like in quantum computing, depends not on the machine but on the human being!

OpenAI has taken measures to control access to its research and Artificial Intelligence technologies. The company made this decision because it realized that some of its technologies could be used in a harmful or malicious way, and of course, to not lose its competitive advantage.

Co-founder and CEO Sam Altman stated that it was a mistake to open-source the GPT code and that they will not open-source their models or share details about how they were trained.

OpenAI may have closed itself, but the war is far from over!

I understand the reasons stated above, but this is a dangerous development, as the progress we have seen in Artificial Intelligence is the result of the open-source mentality. All generations of GPT are based on the Transformer, which was introduced by Google researchers in 2017, so it’s frustrating to see GPT-4 closed off by the company that has benefited the most from open-source.

This is a move back to the era of “trade secrets,” which will stifle innovation and progress in Artificial Intelligence. When I say it will stifle innovation, I don’t mean that it will actually happen, as these few months of “AI Buzz” have already led to the awakening of both “Goliaths” and “Davids,” who can unveil what they have in the forge in record time.

The Age of Artificial Intelligence has begun! Just like the gold rush, all eyes are now on this topic. The beloved Metaverse has taken a backseat, with Mark Zuckerberg’s failed “childish metaverse” plan causing investors who only saw and demanded Metaverse in everything to adjust their sails on their yachts!

With an incredible 100 trillion learning parameters—Machine Learning—the recently launched GPT-4 will forever change the way we think about Artificial Intelligence.

The GPT is an example of a new generation of Deep Learning algorithms that can learn and generate information without human intervention. This is possible thanks to the vast amount of data available on the Internet, which allows the GPT to be trained on a wide range of texts and thus acquire a more advanced semantic understanding.

However, the existence of GPT and other AI technologies raise important questions about the future of humanity. Some of these questions involve the possibility that AI will surpass human intelligence and begin making autonomous decisions, which could lead to a range of ethical and social problems.

For example, imagine a war automaton that can make decisions independently. If that machine decides to attack a target, who would be responsible for the consequences? The algorithm programmer, the automaton operator, or the automaton itself? These are questions that need to be addressed before the technology advances even further.

There are also concerns about the veracity of the information generated by these language models. While GPT’s ability to generate natural language is impressive, it’s still a machine and as such, can be programmed to generate false or manipulated information. This could lead to an increase in the spread of fake news and misleading information, which could have serious consequences for society.

However, it is important to remember that Artificial Intelligence also brings many benefits. The technology can be used to improve efficiency and accuracy in areas such as medicine, scientific research, and natural disaster prediction. In addition, AI can help solve global problems such as climate change and resource scarcity.

To ensure that Artificial Intelligence is used for the good of humanity, it’s important for governments, businesses, and organizations to work together to establish clear guidelines and regulations. This includes creating laws and ethical norms to govern the use of technology, as well as promoting research and studies to better understand the impact of AI on society.

In short, the GPT and other AI technologies are an increasingly present reality in our lives and should be taken very seriously! The GPT-4 is just a step towards OpenAI’s goal of eventually building Artificial General Intelligence (AGI), which is when AI crosses the powerful threshold where AI systems surpass human intelligence.

Man’s reach exceeds his own understanding!

© 2023 Ӈ

Article originally published on Link to Leaders under the Portuguese title: “Explosão de Inteligência — Parte 1÷3”