The final session at CloudWorld 2024 was a culmination of groundbreaking discussions on AI, algorithms, and the Oracle Vector Database, where I had the privilege of sharing my insights on the economic potential of leveraging company data through innovative technologies.
The last session at CloudWorld 2024 was my own on AI, Algorithms, and the Oracle Vector Database. I started with the economic potential for a company based on using your company data (leveraging AI). I showed how McKinsey and Company showed that 25% of a firm's capital is digital capital. The downside was that it could be copied perfectly and infinitely, so your security better be good (why Oracle wins a lot as the solution of choice), and you better not share that information with an LLM outside your company. You need to use that information to innovate better for your customers. But innovation is more than just an attitude; it's moving from current to emerging technology.
I showed how Oracle's autonomous database is self-managing, self-securing, self-tuning, self-recovering, self-scaling, and automates administration by fully automating patching, upgrades, backups, and availability. It puts a patch on your database before you know you need one. Using Oracle machine learning, you can build AI Notebooks in Python, R, or SQL - PL/SQL. My session went through some of the Oracle algorithms (described in Juan's session) that are out there, with many examples of where to apply them. I also showed how you leverage physical devices, like a watch or a phone, that do sensor feature extraction (to see if you're running, walking, or golfing). You can also put those into a grocery app to show people specific pricing (available only in your phone or grocery cart app) or see where people stop within a given aisle (sensors on the grocery cart).
The key to machine learning is first to know the business problem to solve; next, you want to find what function you'd like to perform, such as classification, clustering...etc. Maybe you want to classify good and bad customers, cluster new potential customers into age groups from big data, do a regression to see if you're going to hit your numbers, do anomaly detection to see if there's fraud, or want to attribute importance to see what attributes are making up your good or bad customers. You train the model with 60% of your data, then test and score the model with the other 40% and compare the algorithms used to find which one(s) works best. Oracle also has Auto Machine Learning (AutoML). I used this with a notebook I created (almost 100 pages long) that took a few days. AutoML built the notebook in four minutes. I also found something worth noting at Kaggle (a test data provider) that showed Wal-Mart examples of how they look at sales based on the temperature outside, the gas prices, unemployment, and other things that you usually wouldn't look at for a given business. Once you dive into AI, you'll find there's much more to leverage the better you get at it.
Combining all the algorithms that Oracle has and using GenAI and RAG (to protect your data) will eventually lead to a foundation model for your business (See the well-known Stanford paper on this; I adapted some of the images for Oracle). All of the algorithms are used for business functions needed, carefully weaved with Chatbots that access vectorized data and images to search your database using RAG (to protect your data). Once that Foundation Model is built, it can even learn from other parts of the business and share best practices (transfer learning). You can see Oracle, what they're doing in healthcare right now, and the future foundation model they're building using their own software (APEX / Converged Database / AI) and hardware (Exadata in the Cloud). I think the work in Oracle Healthcare (Cerner Apps) accelerates Oracle's offerings since they're using and improving their own software, hardware, and cloud.
Lastly, I talked about how great Oracle security is and how they have been putting security in Oracle products since 1977; their first customer was the CIA. These include virtual private databases, encryption, label security, fine-grain auditing, transparent data encryption, database vaults, multi-tenant security, Oracle Data Safe, and an in-database SQL firewall (blocking unauthorized SQL and SQL injection attacks). In 23ai, they're looking to add Biometrics and ZPR (zero-trust packet routing).
Think about Star Trek and how advanced it used to look when you were younger; everything in Star Trek has either been invented already, or it's in the process of being completed (other than the USS Enterprise). We're now looking at things, including implants, and how we can leverage technology (especially AI) to accelerate everything in the world with 64-bit technology. We now have quantum computers in their earliest stages. If we look at Google's quantum computer in 2019 and compare it to the one they have in 2024, it took something that would take approximately 47 years to complete in 2019 to just 6 seconds in 2024. That's 241 million times faster in just 5 years!
Oracle also just released the first Zettascale (2.4 ZettaFLOPS) cloud computing cluster (Zetta is 1000x Exa) with 131,072 Nvidia Blackwell GPUs to enable customers to build and train different AI models at scale in the Oracle Cloud. We're moving from using to wearing digital to implanting digital into the hive mind (Gerd), perhaps with Elon Musk's neural network.
In summary, this was a taste of what I saw at CloudWorld 2024. In this blog series, we looked at Larry Ellison's key announcements, which included multi-cloud and incredible security advancements. Larry also said that Autonomous Database and APEX on the 23ai Converged Database would be the key to ALL future and current Oracle products. We saw how Safra Catz ensured companies knew that Oracle was shifting to a customer-first model. She showed that even the CIA was a vital part of the Oracle security solution. Safra stressed that speed and security were Oracle's most significant advantages. Andy Mendelsohn focused on the benefits of the 23ai Database and showed us how to create an Autonomous Database on Azure. Juan Loaiza showed us how to generate entire applications with GenDev using OracleAI and APEX. Steve Miranda showed 100+ Generative AI Features that are already in Fusion Applications. We also saw in my own session where I showed the advantage you can get by leveraging digital. A McKinsey study shows that a company is worth 25% more when it leverages its data assets. I also showed that robots and ChatGPT are already here to help you build your company's future Foundation Model over the next 5-10 years.
Abraham Lincoln said it well, "Things may come to those who wait, but only the things left by those who hustle." If you don't hustle, AI will pass you by. Remember that you're one of the leaders and that 98% of the Fortune 500 run the Oracle database; this new Oracle acceleration speaks to why. Lastly, I want to say there is a safe harbor statement from Oracle that you can see in the image. See me on LinkedIn or X, or email me, but remember all of the great Oracle ACE content out there. See the picture at the end of this article with many of those great Oracle ACEs who take the time to share their knowledge with others. The other secret Oracle has for AI in addition to speed is security, and it's had the best database security on the planet for over 40+ years!
Learn from the ACEs of Viscosity and others out there!