They’re referring to the “AI effect”, where every new development in (narrow) AI causes an equal and opposite shifting of the goalposts. https://en.m.wikipedia.org/wiki/AI_effect

Many, many human-level benchmarks on narrow AI tasks have been passed in recent years. https://www.eff.org/ai/metrics

So, the “well that’s not intelligence, it’s just computation” arguments can continue until it reaches “well that system isn’t conscious, so it’s not AI”. That argument holds no water, since intelligence != consciousness.