Searches for young gas giant planets at wide separations have so far focused on techniques appropriate for compact (Jupiter sized) planets. Here we point out that protoplanets born through Gravitational Instability (GI) may remain in an initial pre-collapse phase for as long as the first $ 10^5-10^7$ years after formation. These objects are hundreds of times larger than Jupiter and their atmospheres are too cold ($T\sim$ tens of K) to emit in the NIR or H$\alpha$ via accretion shocks. However, it is possible that their dust emission can be detected with ALMA, even around Class I and II protoplanetary discs. In this paper we produce synthetic observations of these protoplanets. We find that making a detection in a disc at 140 parsecs would require a few hundred minutes of ALMA band 6 observation time. Protoplanets with masses of 3-5 $M_J$ have the highest chance of being detected; less massive objects require unreasonably long observation times (1000 minutes) while more massive ones collapse into giant planets before $10^5$ years. We propose that high resolution surveys of young ($10^5-10^6$ years), massive and face on discs offer the best chance for observing protoplanets. Such a detection would help to place constraints on the protoplanet mass spectrum, explain the turnover in the occurrence frequency of gas giants with system metallicity and constrain the prevalence of GI as a planet formation mechanism. Consistent lack of detection would be evidence against GI as a common planet formation mechanism.