Hassabis’s Einstein test would be certainly sufficient for the definition of AGI but it doesn’t seem to be necessary. For those who think that it is, Hilbert himself would be a problem to their approach because it’s obvious that Hilbert was a NGI.