this is an excellent post and you raise valid points and considerations. In the past few months we have worked on a few generative ai involved capabilities dealing with natural language processing so I can relate to your post.
1. We started with a hypothesis and built a model that we could play with
2. Then i created an html protototype showcasing the practical application of the model and shopped around customers and internal stakeholders (more internal tbh)
3. we started building and ran into a number of interesting challenges (this is transcription of voice data in public safety) related to domain specific processing of what is said (FPX302 license plate vs what is said over the radio in a phonetic manner foxtrot-papa-etc and a number of other things)
4. furthermore there is processing on top and yes there is latency that we have to overcome
In short yes, the numerous desirability and feasibility concerns have to be evaluated throughout different phases of the project
this is an excellent post and you raise valid points and considerations. In the past few months we have worked on a few generative ai involved capabilities dealing with natural language processing so I can relate to your post.
1. We started with a hypothesis and built a model that we could play with
2. Then i created an html protototype showcasing the practical application of the model and shopped around customers and internal stakeholders (more internal tbh)
3. we started building and ran into a number of interesting challenges (this is transcription of voice data in public safety) related to domain specific processing of what is said (FPX302 license plate vs what is said over the radio in a phonetic manner foxtrot-papa-etc and a number of other things)
4. furthermore there is processing on top and yes there is latency that we have to overcome
In short yes, the numerous desirability and feasibility concerns have to be evaluated throughout different phases of the project
This is excellent insight Dev!