AI development can be split into three categories: developing an ML model (where Java isn't competitive and is unlikely to become top of the class any time soon), developing an AI-centered product (where Java is well-positioned and will become stronger soon; but does this category matter in the long run?) and adding AI-based features to larger projects (where Java is already very good and will only become stronger thanks to Valhalla's value types, Panama's FFM and vector APIs, and Babylon's code reflection).
*Chapters*
0:00 Intro
1:06 Three Kinds of AI
MKBHD - AI the Product vs AI the Feature:
https://www.youtube.com/watch?v=sDIi95CqTiM
2:48 AI Features in Java
TornadoVM: https://www.tornadovm.org/
ONNX Runtime: https://onnxruntime.ai/
DJL: https://djl.ai/
Tribuo: https://tribuo.org/
LangChain4j: https://docs.langchain4j.dev/
Project Valhalla: https://openjdk.org/projects/valhalla/
Project Panama: https://openjdk.org/projects/panama/
vector API: https://openjdk.org/jeps/469
Project Babylon: https://openjdk.org/projects/babylon/
Babylon and Triton: https://openjdk.org/projects/babylon/articles/triton
more on data-centric applications: https://inside.java/2024/05/23/dop-v1-1-introduction/
6:45 AI Products in Java
7:41 AI Development in Java
9:38 Outro
*Tags* #Java #OpenJDK #AI