Yeah, not really.
They decided that they only had to fool 30% of the "judges" into thinking that they were "conversing" with a 13-year-old, ESL retard. They barely made that dubious goal.
Other details are sketchy and the methodology is not scalable (ir)regardless.
This is just typical publicity/funding whoring, neither science nor the can't-get-here-soon-enough, robot apocalypse.