De-extinction is where scientists use methods to bring extinct animals like the woolly mammoth back to life. Currently, no animals have been fully de-extinct as they are only a hybrid of an extinct animal with their closest modern relative. The closest example of de-extinction is the Colossal Biosciences dire wolf project, where they produced a gene-edited gray wolf designed to resemble certain dire wolf characteristics. But still, we do not have one true case of de-extinction unless you count the Pyrenean ibex, which only lived for a few minutes.

According to the literature, there are 7 animals we could possibly bring back. The Dire wolf, which went extinct 10-13 thousand years ago, the Passenger Pigeon, which went extinct in 1916, the Woolly Mammoth extinct 4k years ago; the Quagga in 1883; the Thylacine in 1936; the dodo bird, which went extinct around the year 1662, and the aurochs in the year 1627. All seven of these animals went extinct before 1 million years ago, so we can possibly de-extinct them.

However, de-extinction is still immature and not fully established, and there are some problems. 1. Scientists will use hundreds of millions of dollars just to de-extinct one animal 2. They must have preserved animal bodies for DNA extraction 3. If they lack learned behavior or environment\, the animal that they bring back won't be the same.

So in this project, we hope to solve some of these problems for de-extinction.

We propose of using artificial intelligence (AI) to design the different key features of the animal we are going to de-extinct. AI will get the target traits, like sharp teeth or long tusks, and generate an AI model of it. The model will have AI-designed gene sequences. The advantages are that it takes less time, money, and labor. The disadvantages are that there is a lack of AI models and datasets, but we could use this method later on when de-extinction is more developed. Our method for helping the scientists is to find cheaper and more reliable ways to de-extinct animals so they can bring back some variety to the wild, as many animals that can help us have gone extinct. Some of them, such as the Tyrannosaurus rex, can be detrimental, and we truly hope that we should revive a friendlier type of dinosaur, like the Parksosaurus.

According to the literature, we learned about the existing two common methods of de-extinction. The first method is genetic modification. This method ensures that the animal will have at least 99 percent similarity to the extinct animal. And this method uses somatic cell nuclear transfer (SCNT). Scientists tend to use this method more often as it has more accuracy and therefore is more reliable. But it also has some disadvantages, as this method needs a relative with more than 99 percent DNA similarity, and it can't modify the extinct animal that lived 1 million years ago. The second method is model organism engineering. This method involves adding the traits of an extinct animal to a modern animal with one or more similar physical traits. For example, if we were doing the Woolly Mammoth, we could add the traits to a mouse since both Woolly Mammoth and mouse have long fur. And by doing so, we would get a woolly mouse. The advantages are that we have highly efficient genome engineering, and we also have efficient SCNT and sufficient donor oocytes. But there are also disadvantages, like how it is hard to link extinct traits to the genetic regions of the model organism, and how you make a new animal instead of bringing back one. This is why we proposed our method: using AI for de-extinction.

As an example of using AI for de-extinction, we made a model using Teachable Machine(which uses CNN or Convolutional neural network) that identifies a woolly mammoth skull from its closest relative, the Asian elephant. To develop this model, we found pictures of four subjects. Asian elephant skulls, woolly mammoth skulls, Asian elephant bodies, and preserved woolly mammoth bodies. We used skulls because they were the easiest to identify. In addition, the tusks of the Asian elephant and the Woolly mammoth were of different lengths. Then we used the teachable machine and trained our AI model. Now it can identify a woolly mammoth from its closest relative.

In conclusion, de-extinction is still not fully established. But it has the potential to accomplish many things. It can bring back many fully dead species of animals and add more diversity to the planet. Though not fully awakened to the fullest potential yet, it can be a valuable skill in the future if ever mastered by humans. Currently, not much DNA nor genes have been collected from the extinct animals. At this moment, there is not much hope to revive the dinosaurs, but in the future AI might be able to solve this.

I greatly acknowledge these papers for their information

1. Siqi Wang, Wei Li, Guihai Feng, De-extinction and beyond: trait design powered by generative AI, 2. The Quagga Project 3. About the heath hen 4. BBC Wildlife Magazine 5. National Institutes of Health  6. Teachable machine 7. Oxford Dictionary 8. Live Science 9. ChatGPT for generating images

We want to acknowledge the people who helped us on our journey. In particular, we want to acknowledge our teachers, the school, and the CYSF for giving us this beautiful chance of entering the science fair. And of course, the people who dedicated time to contribute to the sources I am using.