Would starting a PhD involving discriminative vision models still be valuable in 5 years time ?

I am starting my PhD in Computer Science a few months from now. My current research and the work I will be doing in my PhD will be involved with discriminative vision models involving CNNs and Vision Transformers.

But I keep seeing generative models pop up everywhere, including use of generative models for discriminative purposes.

I am worried my PhD work may end up being obsolete and have no value at the end of 5 years, based on how AI/ML …

cnns computer computer science computervision current generative generative models phd research science transformers vision vision models vision transformers will work