I was working on AI in 2018 too :)
At that time, I was quite interested in adversarial examples and ML security.