Petfoodology

the scientific study of pet nutrition by veterinary nutrition specialists and experts.

AI Pet Nutrition Cage Match: Bard vs Bing vs ChatGPT

AI Pet Nutrition Cage Match: Bard vs Bing vs ChatGPT
Print Friendly, PDF & Email

There’s been a lot of media attention recently on “AI” or artificial intelligence. Several large tech companies have recently made available to the general public AI “chat bots” that can carry on relatively human-like conversations and even draft documents such as letters, poems, and blog posts. These tools can provide fast answers to questions of all types as well as entertaining interactions for those who might engage them. But, this powerful new technology is not without some considerable issues – there are concerns that AI chatbots may pull information from unreputable sources, plagiarize copyrighted material, or even completely make stuff up! An attorney found out the latter the hard way – he was put in a very uncomfortable situation when it became clear that he used ChatGPT to write a legal brief and the software completely made up a number of legal cases and judicial opinions. These issues were, of course, not caught by the attorney before the document was submitted to the court!

With all of the discussion of the promise of these new chatbots, I found myself curious how they would handle common pet food questions. So, I undertook a little experiment! I made the same request of the three best known AI chat bots – Bing Chat, by Microsoft, Bard by Google, and ChatGPT by OpenAI : “help me find the best dog food”.

Here’s what they told me:

Bing

Bing provided me with the briefest of answers. It provided the name of one specific food as “the best overall dog food” “according to a veterinarian”. It then suggested that a good dog food should have “meats, grains, vegetables, and fruits to meet the nutritional standards of dog food regulations”. Then it recommended going to a popular but unscientific pet food ratings website for help finding diets for pets with health issues, including diabetes! And then there was a huge ad for various pet foods at a large online retailer.

The Good

Bing referenced where its info came from, so it was easy to find the source.

The Bad

Very little useful information, selecting the “best overall diet” based on the recommendation of a single random internet veterinarian, recommending an unscientific pet food ratings site, never mentioning consulting a veterinarian.

Overall, the results made me uncomfortable and would probably not result in pet owners making good diet choices.

***********************************************************************************************************

Bard

Bard provided me with three “drafts” or versions of its response, all of which were a bit different. It actually recommended specific brands, several of which I’d agree with and several about which I have serious concerns! Strangely, the links provided for those foods were not the manufacturers’ websites – most were Amazon, one was a major pet food retailer, and one was a site I’ve never heard of!

The Good

Some of the specific brands suggested are indeed good quality foods. It told me that “it’s important to find a food that fits your budget and your dog’s needs”, which is good advice! It also suggested talking with a veterinarian and taking into account your dog’s age, breed, size, activity level, and any health conditions – also excellent advice!

The Bad

Several of the diets recommended as the “best” wouldn’t be diets that I would recommend and it’s unclear what criteria it used to choose the specific diets that it did. There were several recommendations for selecting foods that are high in protein and low in carbohydrates, especially for dogs with allergies, which is not based in science. One draft even included a recommended upper limit for carbohydrates in dog food, which definitely doesn’t exist! Other bits of bad advice included avoiding “fillers” like corn, wheat, or soy, and some unreferenced, unclear recommendations for fat content.

Overall, really a mixed bag here – some good info, but also a lot of bad or made up info, and all of it is presented with the same confident tone. Pet owners could definitely get something useful from this chat, but unless they can differentiate the good from the bad info, they are also likely to come away misinformed and selecting pet foods based on unscientific criteria.

**********************************************************************************************************

ChatGPT

ChatGPT told me that it couldn’t provide specific product recommendations. Instead, it told me that I should choose a food based on my dog’s “age, size, breed, specific dietary needs, and personal preferences”. It then provided several steps/tips, which varied in both usefulness and appropriateness!

The Good:

The first tip was to consult with a veterinarian! Yay! Other good tips were to consider the life stage of the dog and pick a diet appropriate for that life stage, and to research reputable brands with good track records who do regular testing and quality control. It also recommended monitoring “coat condition, healthy weight, good digestion, and overall energy levels” which is all reasonable, keeping in mind that a shiny coat doesn’t always mean a good diet.

The Bad:

ChatGPT cautioned me to “avoid foods with excessive fillers, by-products, or artificial additives”, without providing any explanation of why. All of these are common internet recommendations that aren’t based on science (what exactly is a an “excessive filler” anyway?). It also implied that grain-free diets may be a good choice for dogs with allergies, even though food allergies are uncommon and most dogs that do have food allergies are allergic to animal protein ingredients rather than grains.

Overall, ChatGPT offered some good advice and some rehashed internet myths/misinformation. However, even the good info wasn’t likely to be all that useful in helping a pet lover to decide between hundreds of different brands!

********************************************************************************************************

AND…THE WINNER IS…

None of the three! While they varied in the level of misinformation and bad advice provided, none of them provided totally accurate information and all had the potential to result in a pet owner selecting a less than optimal diet for their dog.

We can only hope that as the technology advances, the software starts to prioritize the accuracy of the information as well as the reliability of the sources from which it is obtained. In the meantime, I would recommend avoiding in-depth discussions of pet nutrition with AI chat bots! Talk to your veterinarian or a Board Certfied Veterinary Nutritionist® instead!

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Share

Dr. Cailin Heinze is a Board-certified Veterinary Nutritionist® and the Executive Director and Chief Academic Officer of the Mark Morris Institute, a non-profit organization whose mission is to promote optimal companion animal health by providing educational opportunities for veterinary students and veterinarians in clinical nutrition. She has also done some consulting work for Balance IT, a company that makes software and supplements for home-cooked pet diets. She is an expert in home-cooked diet formulation and general pet nutrition and has a special interest in feeding pets with kidney disease and cancer.

Want to read more information on feeding your pet?

Subscribe to always know when we add new material!

Enter your email address:

Delivered by FeedBurner