The original mistake
The assumption that birds have poor olfaction came from anatomical comparison: bird olfactory bulbs are relatively small compared to mammalian olfactory bulbs at equivalent body sizes. The inference — that small bulbs mean weak smell — turned out to be a flawed proxy. The function of an olfactory system depends on more than bulb-to-body-size ratio, and birds have evolved olfactory architectures that are different from mammals rather than simply scaled-down versions of them. Across the 1990s and 2000s, accumulating behavioral evidence and updated neuroanatomy showed that the original conclusion was significantly wrong for many bird groups.
Several documented or strongly-suggested olfactory uses in corvids.
What corvids do with smell
Several documented or strongly-suggested olfactory uses in corvids. Cache recovery may incorporate olfactory cues alongside spatial memory, particularly for refining the exact location of buried food items. Scavenging behavior, which is a significant part of the corvid foraging niche, plausibly draws on smell to locate carcasses at distance. Conspecific recognition may have an olfactory component, though this is less well-studied than the vocal and visual components. Predator detection in some scenarios may rely on smell, particularly for mammalian predators with characteristic scent signatures. None of these uses are as dominant as olfaction in dogs or rodents, but they aren't negligible either.
What the research actually shows
The systematic experimental work on corvid olfaction is relatively recent and limited compared to the visual and vocal literature. Several studies on common ravens and carrion crows have documented behavioral responses to scent cues in controlled tests. The Australian magpie (a corvid relative, though not a crow) is one of the better-studied species for olfactory behavior. The general pattern: corvids respond to olfactory cues in detectable, measurable ways, but the response magnitudes are smaller than for visual or auditory cues. Smell is a real sensory mode for these birds but not the dominant one. The honest framing is 'birds smell more than we thought, but not as much as a dog,' not 'birds are great smellers.'
Why this matters for communication research
Crow communication is overwhelmingly studied as vocal communication. The acoustic dimension dominates the research literature, both because vocalizations are the most-prominent signal type and because microphones are easier to deploy than systematic olfactory measurement. But the underlying communicative situation may include non-vocal channels that get systematically under-represented in the data. If a crow is responding to a conspecific in a complex social context, the response is informed by what it hears, what it sees, and potentially what it smells. The acoustic record captures one channel of a multi-channel signal exchange. Acknowledging this is part of being honest about what bioacoustic atlases can and can't tell us.
Why nobody's building 'crow olfaction AI'
Practical reasons. Microphones are cheap, ubiquitous, and produce digital data that machine-learning models can consume directly. Olfactory sensors at chemical sensitivity comparable to bird noses are emerging research instruments, not commodity hardware. The species-relevant scent chemistry is much less mapped than the species-relevant acoustic patterns. And the experimental setups for testing olfactory hypotheses are harder to design than playback experiments with sound. The field will probably get there eventually — electronic-nose research is advancing — but for the next decade, AI bioacoustics will continue to be where the data is, not because acoustics is the only or even the most important channel, but because it's the one we can practically measure at scale.