If you view YouTube at all, you see that the site is plagued by false news reports that are generated by AI. The problem with AI is the people who program the software. AI made it easy to research documents, but AI is not without bias. AI is slowly changing the way we think about things and is capable of changing our “real” history to a false one. Every time I see a data center being built, I worry about how our knowledge will change for the worse. How will they invade our lives? How will they move us to a false reality? How much will my spam increase? How much data space will be given to hackers?
Best Friends recently had a conference geared towards using AI to get pets adopted from shelters. That seems like a noble notion. During my career, I was constantly fighting volunteer groups who wanted me to provide a false narrative of the animals in my care. I demanded the integrity of my organization. I wasn’t well-liked. I worry about how shelters filled to the brim with Pitbull dogs will use AI to represent the dogs as another breed. I saw that the new software will even write up the animal’s description for shelter staff. How valuable is integrity to this AI software? It is only as good as its programmers.
Is the intent of the software to move animals out of the shelter, or is it intended to place animals into the right homes? With computers, input equals output. Good input equals good output. If the input doesn’t include the families that will be adopting these animals, then we will see a rush to adopt animals and then later a rush to abandon them (again).
Let’s face it, the No Kill Movement is a statistical movement. Everyone wants to be a no-kill shelter, and all they have to do is adopt 90% of their animals. Even if by hook or by crook. Is AI a crook? It is everywhere I see it being used. I get around 100 emails daily that are generated by AI. Fortunately, I have a fairly good spam filter. If you have ever used a dating app, you have experienced disappointment. Don’t get caught up in the fad of this type of placement; do your homework. If the pet that you are looking at had a previous owner, ask the shelter staff to pass along your phone number to them so that you can inquire as to the time that they spent with the dog.
The problem with the No Kill Movement is that a shelter’s statistics is more important than a good adoption. The best way to solve this problem is to leave a review on Google Maps about your adoption experience a few months after your adoption. A few bad reviews might cause a shelter to more accurately describe their animals in the future.
I hope that AI will be used as a positive means to place pets into loving homes. But the past use of AI should make everyone skeptical. Of course, any shelter that has lied about their animals in the past will become better liars with AI.
I think AI is best used for sifting through data. If an animal shelter has a large database, AI would be useful when lost animals are found without identification. AI could search the shelter’s database to see if the animal had at any time previously visited the shelter. Most animal shelter databases do not make use of AI. Shelter employees are forced to conduct manual queries of their data in hopes of finding a hit. Of course, that would mean that shelter employees would have to go into greater detail when describing the animals on intake. For example, the animal’s left paw is black. We don’t usually list a specific descriptor for the left paw.
Our attempts at trying to create a complete software package led to the death of PetWhere. We attempted to do too much. And at that time, shelters were asking us to make it simpler. If we had listened to the animal shelters, PetWhere might still be alive today, but half of the animals in the shelter would have a breed designation of “mixed;” making the software useless.
However, an optional description field could list a black paw or list the various breeds that we can identify in an animal. The earlier databases were preferred by lazy staff who just wanted to list the breed as “mixed.” As with many shelters today, wanting to eliminate Pitbull as an animal descriptor. AI would help staff identify an animal without using the word “Pitbull.” AI could also be used to take the photos of Pitbulls and make them look less Pitbull on the shelter’s website.
I think AI’s best use in shelters would be to take an animal’s intake photo and compare the photo to animals who have previously been in the shelter. I can recall incidents in which owners attempted to misidentify an animal on reclaim to avoid a higher impound fee. But, that only works if you haven’t doctored the animal’s photo to something less than a Pitbull.
Or a better use of AI is the ability to take a picture of an animal and report back the various breeds it may belong to. You could fine-tune that with DNA testing. When I first started in the animal welfare business, we could take a universal breed book and decyhper 99% of the animals. You can’t do that anymore. Breeders have gone nuts in interbreeding dogs. The idiot breeders bred Pitbulls. As a result, most shelters have a population of 70% or higher of Pitbulls and Pitbull mixes. The only real solution to shelter overcrowding is to eliminate Pitbull breeding.
AI is brilliant in its ability to deceive as well as its ability to sift through disconnected data. Just as the no-kill movement is a good desire, if it can be done honestly.