The second article is marginally more on point - 24% fewer crashes for vehicles with lane keeping assist (so my guess at the meaning of the 23% stat may have been wrong). But the 95pct confidence interval is 2-42% and the study acknowledges that its efforts at controlling for confounding factors in the type of cars that have this feature are imperfect. It also took place in the US, so there's certainly no mandate for haptic feedback and I suspect very few cars had it. This is marginally more helpful evidence but not very good, I think--it seems very plausible that audible lane keeping features are helpful and moving your steering wheel (which sounds terrifying) is unhelpful.

As an anecdote, I crashed a car as a teenager thanks in part to panicking (unnecessarily) when a rough highway started moving the car's wheels (which I noticed of course via the steering wheel) without my intending it. Fortunately there were no injuries.