The difficult and limiting part of scienctific work isn't coming up with ideas (which is, at best, all a glorified Montecarlo chain autocorrect can do) but actu

The difficult and limiting part of scienctific work isn't coming up with ideas (which is, at best, all a glorified Montecarlo chain autocorrect can do) but actually designing and performing good experiments. Add to that the amount of junk papers that will be generated and published - through which all future scientists will have to sift - and AI in it's current form is much more likely to be a concrete block dragging scientific progress down.