In Allen Zhu’s recently speech advising startups to lay off people early and decisively, he also mentioned rat-catching AI as an example of how F&B outlets are automating in the aftermath of Covid-19 outbreak. 
This sounds a bit puzzling to me – how can AI catch rats, especially from a technological point of view?
Does that mean a much more advanced, much nimbler and of course much cheaper version of Boston Dynamics WildCat?
Or a more ferocious Doraemon?
Well, in fact, after checking with some friends, I realised it actually referred to rat detection using image recognition. Borderline AI.
They do not have a solution of actually acting on such detection (smart mouse trap, laser gun etc.), yet.
Actually, AI solutions are almost exclusively 2B now. If you have the right relationships and right connections, rats? You can catch bad guys too!
Public security has been the largest AI use case (and reason for procurement) thus far. The estate where I live recently installed an automatic infrared body temperature detection device. Imagine the procurement amount if you put one such machine in every neighbourhood in Beijing.
However, the problem is, my temperature for the last month has been constant at 34.1 degrees celsius. So is the body temperature of all delivery riders who send me meals and parcels.
Back to rat catching – a more useful use case, which a few friends have requested from me, is mosquito-catching. Anyone who lived on a low floor apartment in Beijing summer would know what I am talking about.
But similar to rat catching – AI does not have a carrier. Unless you build a robot gecko…
Or as some people suggested, use auto laser shooters.
But, given the amount of mosquitoes in Beijing summer, your balcony would probably look like this:
