Appleapplemon2

AI will just end up being API calls within CRUD apps

All this AI hype but when it settles, all that will remain are AI libraries with APIs that other programs use. Same like the last 20 years. Most won't need to know internal implemention details.

Poll
57 Participants
Select only one answer
AMD quantumai May 16

It has to be like api calls , as most of the people who will use won't understand the mathematics behind it People who will understand mathematics may not understand the software written to train it. People who will understand software won't understand the hardware underneath it. Also for traditional computing full stack debug till hardware level was possible. For ai it won't be as it's impossible to know which exact functions are mapped to which instance and what operation as the network decides the flow of data . Hence it will be more obfuscied, if there are no API calls using them is almost impossible for most of the workforce.

Amazon smolfriend May 17

isn’t the latency way too large? they will get creamed by on device models that make use of greater efficiency and innovations in special AI hardware. Networking hardware probably won’t improve much

AMD quantumai May 17

Yes there will specialise inference/ai engines, mainly vector processors to accelerate/run the ai intensive tasks. Just the architecture will improve over time, the compiler and software will remain the same. It's just that people will have to learn to get used to it.

Edelman Financial Engines kevinchili May 17

Yep. But for sure the hiring managers will ask you about it because there’s plenty of people in the market.