OpenAI CEO, Sam Altman, is excited about getting more investors to back up the creation of the next-generation AI model. The tech guru is on a mission to create Artificial General Intelligence (AGI), a software as intelligent as human beings. Sam is banking on his biggest investor, Microsoft, to pump billions of dollars into the future-altering project.
Sam Altman and Microsoft CEO, Satya Nadella have a partnership that is “working really well.” According to reports, Microsoft invested $10 billion in OpenAI earlier this year. Altman revealed that he is expecting more from Microsoft and other investors. Adding that building a more sophisticated AI model is very costly.
Reports allege that the San Francisco-based company, OpenAI, has a valuation of $29 billion. According to Sam, who refrained from releasing any figures, revenue growth from the company has been good this year. However, the company remains unprofitable due to training costs. The expenses incurred in training models are punishingly huge for the tech guru.
Nonetheless, a Microsoft investment would keep things running smoothly hence ensuring that both sides make money on each other’s success. Sam revealed plans to make Chat GPT a business franchise. These plans include existing tools such as custom versions of Chat GPT which can be modified and fitted for specific applications. Additionally, there is a GPT Store, which will operate in a business style similar to Apple’s App Store.
Despite having a research lab, API software, Chat GPT, GPT Store, and a partnership with Microsoft, Open AI’s sole product is Intelligence. Sam noted that “magic intelligence in the sky” is what the company is all about. To build the company, the CEO hired renowned experts such as Dropbox’s and Y Combinator’s COO, Brad Lightcap.
Sam Altman is currently focusing only on two things, building a superintelligence and building the computing power to do so. These tasks not only involve making the AGI but also making it safe and figuring out its benefits. OpenAI is also looking to build more ChatGPTs (autonomous agents) that are more powerful and require more complex actions.
As per Sam, these new autonomous agents will be able to execute codes, make payments, send emails, and even file claims. Moreover, the business value they promise to bring is “pretty good.” However, Sam did not mention when Chat GPT-5 will be released. He noted that this next-generation AI model requires more data to train on. OpenAI collects this data from data sets made public on the internet as well as proprietary data from companies.
Earlier on, OpenAI had called upon organizations “not already accessible online to the public” to provide it with large-scale data sets. These data sets were specially for long-form writing or conversations in any format. Unlike its predecessors, GPT-5 is likely to be more advanced. Hence, predicting the new capabilities and skills it might hold is hard at the moment. Until the model is trained, trusted, approved, and tested, speculating would be like a “fun guessing game.”
OpenAI uses advanced H100 chips from Nvidia in its technology. Over the past year, rival tech companies have rushed to obtain these semiconductors from Nvidia. As a result, there has been a supply shortage of the $40,000-a-piece chip. However, Altman stated that he received the chips needed to build AI systems and was expecting more soon. Nonetheless, due to its demand, Google, Microsoft, AMD, and Intel are preparing to release rival chips.
Consequently, Nvidia’s chips will be in less demand should rival companies succeed in their mission. Fortunately for OpenAI, it took an early lead in the race to build generative AI systems. Despite proving with GPT-3 that these systems can create text, images, code, and other multimedia in seconds, other tech companies have yet to do it. Nevertheless, progress toward building AGI requires large language models (LLM). These LLMs according to Sam, are one of the core pieces of building AGI coupled with other pieces on top of it.
Subscribe to Switch TV
The tech company has focused mainly on LLMs while its competitors are still scouring alternative research strategies. However, Sam and his team believed that language was a “great way to compress information” therefore developing intelligence. Google DeepMind, as per Altman, missed the fact. Nonetheless, Sam Altman is determined to find the biggest missing piece which is required to take GPT-5 to the next level.
Altman argued that Newton would never have invented calculus by reading only about geometry and algebra, and neither would his models. To him, finding the missing idea to generate new knowledge for humanity is the biggest thing to work on.