I have heard occasional rumors that effective field theories have gravity duals. For example, I've been told that UV momentum cutoffs in N=4 SYM become finite radii in AdS. I've heard speculations about AdS duals of plain old QCD. And I know it's thought that CFTs always have gravity duals.
Is it believed that any UV completion of an effective field theory must have a Maldacena-style gravity dual (in Witten's sense, meaning that currents become boundary conditions)? Can QFTs with gravitational sectors have gravity duals? What about string theory?
Are there any necessary reasonableness conditions? (Maybe insist that the renormalization flow is a gradient flow?) Is locality necessary? (String theory seems to think that gravity must be holographic. Can you have a holographic description of a holographic description?)