I'm curious if anyone here has actually gone through the process of buying components and building their own AI system. I've heard that there are some pre-built algorithms you can start with instead of starting from nothing. Has anyone had experience doing this?
3 Answers
Building AI can be straightforward—like creating your own image classification system. That said, making a large language model (LLM) by yourself is another story; it's quite a tall order!
It really depends on what you mean by AI. True LLMs need enormous amounts of data and computing power to train. For example, one claim for training an LLM was around 5-6 million USD, and that was just for training! The complete development costs can reach billions, involving huge teams. But if we broaden the definition of AI, there are definitely simpler machine learning models you can tackle on your own.
No one has really managed to do it all from scratch yet! However, there are tons of libraries and utilities available that can help you kickstart your machine learning journey. If you're up for a challenge, implementing a simple neural network is definitely within reach for most programmers. But if you’re thinking about a career in machine learning, you'll probably need a degree, and the journey is going to be tough. A good starting point is reading the '100 Page Machine Learning Book' to grasp what you're getting into. It's a lot more complex than it seems—math will be your constant companion!
Related Questions
Biggest Problem With Suno AI Audio
How to Build a Custom GPT Journalist That Posts Directly to WordPress