A unique attempt to implement neural networks, artificial intelligence that mimics the human brain, using SQL, a database management language, is gaining attention.
Imagine this: What if the rigid Excel-like database you use daily for managing company sales or inventory suddenly started telling you, ‘Boss, we expect sales to be this much tomorrow,’ or accurately predicting customer preferences? Typically, we think we need to extract data from a database and move it to a specialized programming environment like Python to use AI. However, a fascinating challenge is emerging among developers: ‘What if we didn’t move the data, but ran AI directly where the data is stored (in SQL)?’
Why is This Important?
Embedding AI at the point where data flows is akin to ‘completing quality control while manufacturing products in a factory.’ Normally, extracting data from a database and sending it to an external AI model incurs time and costs. However, by performing predictions directly within an environment like SQL Server, we can reduce complex data movement processes and analyze data more quickly and efficiently Source: SQL Server Downloads – SQL Server Central.
In our daily lives, this is like the convenience of a smartphone photo app instantly finding people’s photos within the phone itself, without needing to connect to a cloud server. Since there’s no need to pull data out, speed increases, and security concerns are significantly reduced.
Easy Explanation: A Neural Network is a ‘Mesh of Little Filters’
So, what exactly is this ‘Neural Network’ running in SQL? It might sound difficult as a technical term, but a simple analogy is that a neural network is like ‘tens of thousands of little filters that exchange information and learn from each other.’
- Connection of Neurons: A neural network consists of simple units called ‘neurons’ densely interconnected like a net. They send signals to each other to perform very complex tasks Source: Neural network - Wikipedia.
- Brain-like Structure: This structure draws inspiration from how the human brain processes information Source: Neural Network from Scratch. In this article I’ll implement a neural. Just as multiple parts of our brain react simultaneously when we see an object and recognize it as ‘an apple,’ neural networks also solve problems through the collective effort of neurons.
-
Weights and Layers: A neural network is formed by stacking simple neurons layer by layer. When data is input, it learns using the ‘weights (importance)’ and ‘biases (thresholds)’ of each neuron. Simply put, as information passes through, the little filters refine, filter, and learn from each piece of information, ultimately deriving the result of ‘What is this?’ [Source: What Is a Neural Network? IBM](https://www.ibm.com/think/topics/neural-networks).
The goal is to implement this complex process using SQL, the language we typically use for data organization. It goes beyond simply using Excel functions for basic calculations, aiming to make the database itself capable of reading patterns from the data.
Current Situation
| Currently, many developers are building AI’s foundational strength by implementing neural networks from scratch. Practical exercises for implementing neural networks in various environments are already actively underway [Source: ShowHN: I implemented a RNN from scratch by… | Hacker News](https://news.ycombinator.com/item?id=44879741). Of course, we cannot directly insert massive and complex models like ChatGPT, which we use daily, entirely into a database. However, as database experts suggest, the technology to embed basic and simple predictive models within the database is gradually establishing itself in practical fields Source: SQL Server Downloads – SQL Server Central. |
What Happens Next?
In the future, database administrators might evolve from mere data organizers to ‘AI cultivators within the database.’ This is because the future of data management lies in gaining immediate insights from the safest and deepest repository where data resides. The systems you use will also eventually learn from data quietly in the background, providing smarter answers.
MindTickleBytes AI Reporter’s Perspective
If databases, the traditional repositories, also take on the role of AI brains, the bottlenecks caused by data movement will disappear. The combination of the classic tool SQL with modern neural network technology is a prime example showcasing how AI can integrate closer to us and become a more natural presence. An era where databases themselves become ‘smarter’ without relying on complex external AI models is rapidly approaching.
References
- Neural network - Wikipedia
- SQL Server Downloads – SQL Server Central
-
[ShowHN: I implemented a RNN from scratch by… Hacker News](https://news.ycombinator.com/item?id=44879741) - Neural Network from Scratch. In this article I’ll implement a neural
-
[What Is a Neural Network? IBM](https://www.ibm.com/think/topics/neural-networks)
){kind=link}
- Transistor
- Neuron
- Database row
- Data compression
- Predictive analytics
- Improving web search speed
- Computer memory structure
- The structure of the human brain
- Communication network routing methods