A stupid observation: I’m struggling to see how AI companies get their money back.
The cost/benefit is already too hard to justify. Maybe the benefit is hard to measure. Maybe the benefit is too low.
So either:
– existing models need to get cheaper to stimulate use
– models will need to get better – but better models cost more to train and use – higher costs harder to justify
– more efficient models come along with lower inference costs and higher margins – is anyone perusing this?