"if you give IBM your data, let them train a model on it, and communicate what you want, you will get an end-to-end custom solution"
If you give a summer intern your data and communicate what you want, they can also give you a custom solution at a fraction of the price. IBM's model might be more advanced, but the real challenge is applying the conclusions, rather than just building a model.
Communicating what you want is the hardest part. Most managers who are excited about AI (including my boss) have no idea what they want to use it for. "Making things more efficient" or "analysing what's going on in the factory" is way too vague.
Getting data is not so hard, but getting good metadata is a challenge. For example, downloading some music in Chinese and tagging the artist/play count/lyrics/etc. When I have all that metadata, I might not need the machine learning part - I can just find the top 1000 songs and process them manually in an afternoon.
Before spending millions on Watson, I think companies would have a higher return on investment by hiring a summer intern to come up with ideas for further research, and gather the data. Then they can apply AI if they really need it.
If you give a summer intern your data and communicate what you want, they can also give you a custom solution at a fraction of the price. IBM's model might be more advanced, but the real challenge is applying the conclusions, rather than just building a model.
Communicating what you want is the hardest part. Most managers who are excited about AI (including my boss) have no idea what they want to use it for. "Making things more efficient" or "analysing what's going on in the factory" is way too vague.
Getting data is not so hard, but getting good metadata is a challenge. For example, downloading some music in Chinese and tagging the artist/play count/lyrics/etc. When I have all that metadata, I might not need the machine learning part - I can just find the top 1000 songs and process them manually in an afternoon.
Before spending millions on Watson, I think companies would have a higher return on investment by hiring a summer intern to come up with ideas for further research, and gather the data. Then they can apply AI if they really need it.