“Leveraging advanced mathematical models in artificial intelligence (AI) and machine learning, we have developed a system capable of interpreting various emotional states in chickens, including hunger, fear, anger, contentment, excitement, and distress,” the researchers wrote.
TLDR: it’s not translating speech as much as it is interpreting tone of speech to understand mood.
From the GinaD on the Earth Species Discord:
"As someone who’s managed a proof-of-concept grant program at a research university (UC Davis, which has a world-class veterinary school), I can say that this paper is just plain bad. It seems to be a term paper that details standard CNN architecture and modeling approach (though the Discrete Fourier Transform was a nice touch) and doesn’t provide the most important information: How many data points were fed into the model, and how on Earth you determined these emotions in chickens. As I’m sure everyone here knows (hence our great mutual interest in the topic of understanding, and potentially communicating with, other species), determining the precise emotional state of animals is not a trivial task.
A few other points on this paper…. For one thing, they tell us nothing about their affiliations or who they are at all. Sloppy. Second of all, any paper that starts off with, “In this groundbreaking study…” is immediately suspect. Third, for a paper that is so sloppy in other respects, the writing / command of the English language is so good (better than a lot of native English speakers) that I can’t help but suspect plagiarism or at least heavy reliance on Chat-GPT or similar. (I ran it through a plagiarism scan and it was flagged for plagiarism, but I would have had to upgrade to a paid account to find out if it was just comparing to the original paper, or if there was actual plagiarism.)
The goal of the paper is laudable, but there’s no way they will ever get to the peer-review phase of any journal with this paper as it is. My guess is that this is a class project (lead author is at iUniversity in Japan) and they published it to a pre-print journal website because … they could.
While the peer-review journal publishing process has its faults, it at least prevents people from publishing so-called “research” that is at best a waste of the reader’s time, and at worst is misleading or downright fraudulent.
I took the time to write all of the above because it’s important to make an effort (as several of you did; thank you for that) to dig into the claims a little bit. Unfortunately, if people don’t critically evaluate and challenge these kinds of claims, they can take hold among a broader group of people and divert attention from the good work being done by careful, thoughtful researchers."
