for practical purposes --------- much AI is already way ahead of most PhD people and some models - way ahead of everyone
LLM's are practically very useful - speeding up research enormously - even with hallucinations - it's their provision of references that is the useful thing atm
Specific models - like some of the Microsoft models used in materials research - are already way way way in front of any research teams
in 5 years - the changes will be stunning, in 10 years - beyond belief
