Generative AI changes everything
Larry Ellison kicked off his keynote by delving into Artificial Intelligence, particularly focusing on ChatGPT. The phrase is sure to become famous “About a year ago OpenAI demonstrated ChatGPT 3.5 and much to the shock of the people who developed ChatGPT… the baby talks. Seriously the developers did not expect that“.
Larry then revisited OpenAI’s history, explaining that it was developed a decade ago. He notablity explained that they constructed an enormous neural network composed of billions of artificial neurons. After building it, they extensively trained it on a vast amount of data, including the entire public Internet. However, the underlying concept remained consistent with previous iterations. What set ChatGPT 3.5 apart was its larger neural network and broader training data. Due to this increase in scale, when prompted with questions, ChatGPT unexpectedly provided answers and engaged in conversation. This revelation surprised the ChatGPT team, AI professionals, and the world at large, capturing our collective imagination.
However, as Larry mentioned, most cool tech doesn’t come without risks. Everyone wants to know what comes next. “What comes next is a wordwide race to build what comes next, to build better AI“. While better features will undoubtedly emerge, they also bring new risks, like any new technology. “From early innovations like fire, which could be misused, to later technologies such as nuclear power, all new technology carries potential for misuse“.
As stated in Larry’s slides: “Is Generative AI the Most Important New Computer Technology Ever? Probably!”
To the question what does the future hold? “Complete self-driving cars within the next 12 months from Tesla, computer-designed antiviral drugs, and much more.“
Oracle Cloud Best Cloud for training Generative AI Models
Larry mentionned that Oracle Gen2 Cloud stands out from other clouds in several significant ways. According to him, one of the key distinctions is the network that Oracle employs to interconnect its computers. He refers here about the RDMA (Remote Direct Memory Access) Network which connects NVIDIA GPUS into Huge Superclusters. This network grants Oracle the ability to transfer large volumes of data between computers at incredible speeds, surpassing conventional networks. This remarkable speed enhances Oracle Cloud’s capacity for rapidly training AI models.
In his own words: “We are much faster and many times less expensive than the other clouds for training AI Models“
Larry Ellison then came back on NVIDIA’s announcement regarding the construction of the world’s fastest AI supercomputer, leveraging the NVIDIA H100 GPU and the Oracle Cloud RDMA Network. He emphasized that while the processing speed of data is crucial, the ability to move data quickly between computers is equally vital; otherwise, processors remain idle. Oracle’s exceptional data transfer speed advantage stems from their early adoption of this technology for their own database technology, effectively solving the data movement challenge.
Generative AI is Transformational
Larry explained that Generative AI is fundamentally changing how Oracle (will) build and run applications. He provided the following examples:
- Improved APEX Application Generator: No Code – not – Low code
- Improved Autonomous Elastic Database: No Human Labor – No Human Error
- New Object Relational Database: Generate Relational Schema from JSON documents
- New Oracle Vector Database: Specializes Generative LLMS for Medicine, Law…
- New Cloud Data Intelligence Platform: Population Scale Data Trains Medical AI Models
Apex for all future Oracle Applications development?
According to Oracle’s CTO and co-founder, most new Oracle applications will now be generated with APEX, including Cerner New Millennium, Fusion Marketing, Banking, Retail, Hospitality, and more. APEX enables development teams to be streamlined, significantly reduces development time, and ensures more secure and reliable results. While existing Java projects at Oracle will continue, no new Oracle products will be developed in Java, despite Oracle’s history of using Java for development.
Additionally, all Oracle new or strategic applications will now run on Autonomous Database, typically Fusion & NetSuite will move to Autonomous Elastic Database. Why this decision?
“If there is no human labor, there is no human error… the only way to build a secure system is to eliminate human error.”
Larry then discussed the utilization of AI in the healthcare sector, including medical AI models, and the methods for collecting patient data using Cloud IoT devices such as blood pressure monitors, pulse oximeters, smartwatches, and more.
AI can transform other industries
Following the previous examples that were given and aside from medicine, there are other industries that can be transformed by Artificial Intelligence. The CTO and co founder provided us with a compelling example from the agriculture sector, specifically Greenhouse Agriculture Robots. These technologies offer various advantages, including:
- Grow using 98% Less water
- Grow using 90% less land
- Grow new Population Centers
- Daily Harvest with Same Day Delivery
- Better Job Security and Job Quality
Other examples were shared, such as the current police cars. This also presented the opportunity to give us a glimpse of the future police cars (Larry and Elon’s personal favorites 😉). A few key points about these cars: “They are safe, exceptionally fast, and have stainless steel bodies.“… well I would say not a surprise.
MultiCloud: Microsoft & NRI
Last topic that Larry addressed was his visit to Redmond last week where he met with Microsoft’s CEO, Satya Nadella. During their meeting, they reached a consensus on significant ideas: “Clouds should be open” and “Clouds should be interconnected“. Many customers currently use multiple cloud services, including AWS, Azure, Google, Salesforce, and Oracle.
Customers should find it easy to utilize various cloud services and seamlessly interconnect them, regardless of the provider. The concept revolves around ensuring that all these clouds can effectively interoperate.
Both clouds are open and both cloud providers would like to avoid Egress fees on customers data. To the question “Is it a complete Oracle Cloud Inside the Azure Datacenter the response is Yes !“