1. It doesn’t exist
What we think of as AI would really be something like general intelligence, that we are nowhere near yet. We can’t even work out how a human brain works so we would be hard put to replicate it.
2. Limited knowledge on application
How do we implement technology and keep users or operators at the heart of it so that decision making has oversight and context.
3. Computing Power
Algorithms are getting more and more complex demanding more and more computing power. In a world driving towards sustainability and limitation of climate change this is ever a concern as to how we grow technology sustainably.
Due to the selling and marketing going on around AI there is a tendency to perceive the technology as something it is not. This causes a trust issue when we find out that the computer can’t actually do all those really fancy things we got shown in the advert.
Data is the basic ingredient to any model but data collection can be somewhat overlooked. If we put rubbish into models we are bound to get rubbish out. An adoption of a modelling methodology is key to ensuring models fit for purpose using robust data, collected in an ethical way.