ANIMAL INTELLIGENCE WITH AI: DECODING NATURE'S SIGNALS

How can AI help us decode nature's signals for a sustainable future? This article explores the delicate balance of ecosystems, lessons from animal behavior, and how AI can be a powerful tool for conservation, while also considering its environmental impact.

•

*This article was originally written by Shree and submitted as a paper for the All India Seminar conducted by the Institution of Engineers (India), BLC on Engineers Day 2024.

ANIMAL INTELLIGENCE WITH AI: DECODING NATURE'S SIGNALS FOR A SUSTAINABLE ENVIRONMENT TOMORROW

How can Artificial Intelligence help us unleash the hidden intelligence of animals, and how can it assist us in decoding nature's signals? How can we leverage this knowledge to benefit ourselves? Why does this matter for sustainability? Additionally, in this article, you will find out some fundamental aspects of AI, including how it works and how LLM like ChatGPT functions and its negative impact on our nature. Let's dive in and explore these important topics. To begin, let's take a closer look at ourselves and our surroundings first.

NATURE'S WARNING SIGNS: FROM OUR BODIES TO OUR PLANET

The human body is a fascinating piece of engineering, created by God. Our bodies are smart; they send us subtle signals when something is wrong - like rashes, hair loss, or pimples. This way, the body is telling us, "Hey, there is something wrong with your gut, or maybe with your liver, you should address that before it's too late." Similarly, our Mother Earth communicates with us through various means, such as natural events like cyclones, floods, droughts, and landslides are all signs that something is wrong in our environment. Often, human activities are the root cause of these disturbances. What are those activities? Let's find out.

THE IMPORTANCE OF BIODIVERSITY

Cutting down too many trees, polluting our water, and installing excessive mobile towers disrupt nature and leave many species homeless. These actions harm animals, aquatic creatures, birds, and even insects by disrupting their natural behaviours and navigation abilities, throwing ecosystems out of balance and disturbing the delicate equilibrium of our environment. Every organism, from the tiniest insect to the largest elephant, plays a crucial role in maintaining the health of our planet. When we harm one species, it can trigger a chain of unexpected consequences that affect the entire ecosystem. Let's explore these consequences and examine our impact on nature.

THE DELICATE EQUILIBRIUM: UNDERSTANDING OUR IMPACT ON NATURE

I was at an event where I listened to speeches by Dr. Santosh Padhi [4], a renowned medical doctor and public health officer from our state of Odisha, who is actively involved with the government in preventing various diseases caused by mosquitoes by introducing innovative solutions. Out of curiosity, I asked him, Sir, why not just eradicate all mosquitoes at once? Why are we not focusing on finding a way to do that? Because, Mosquitoes are responsible for killing lakhs of people every year, while predators like lions and tigers come nowhere close to mosquitoes. He replied, that's precisely what we're working on, but completely eradicating them from the planet is simply not possible. However, even hypothetically, if we could do that, we may never know what would happen to our ecology. Many insects and birds depend on mosquitoes. If we eradicate them, those birds will die, those insects will die, which will affect other species that rely on those insects and birds. We could face a domino effect leading to ecological collapse. Of course, we need to find a balance. If we don't manage the mosquito population, it could grow uncontrollably, which would be detrimental to both us and the surrounding environment. It's crucial that one form of life does not dominate over another. This is why it's so important to protect all forms of life, not just the ones we find cute or useful.

Recently, I had the opportunity to talk to Dr. Mary Trudeau [5], a renowned environmentalist from Canada who works closely with the Canadian government to address issues related to water projects and climate change. She shared an interesting case study with me: the remarkable recovery of the Thames River in the UK, home to iconic landmarks like the London Eye. Today, it is one of the cleanest rivers on our planet, but this was not the case earlier; in fact, it was completely the opposite. The river had become so polluted from urbanization and industrialization that it was declared biologically dead. At first, people thought that simply adding fish back into the river would help. However, that approach didn't work. They then sought to find the root cause of why the fish did not want to live there. As a result, they focused on cleaning up the pollution and improving the water quality. Over time, fish and other wildlife started to return on their own! It was obvious why the Thames River was biologically dead. We see a similar situation in India, where many rivers and lakes are polluted beyond warning levels due to industrial waste and local pollution. The river Ganga, which is sacred to us, faces severe ecological challenges as a result of this pollution. No aquatic creature wants to live or survive in polluted water. But there are other, less obvious issues we humans create that indirectly impact these ecosystems. This story reminded me of something I experienced first-hand.

I remember when I was in kid, I'd see sparrows (ଘରଚଟିଆ) every once in a while. I asked my mother why there were so many more pigeons and crows than sparrows, and she replied that those birds usually stay on hut houses, but nowadays it is difficult for them to find a hut and also a tree. As we continue cutting down forests and building structures, they are very sensitive to pollution, and it is difficult for them to find food too, hence the sparrows are gradually leaving us, along with other birds. If we continue what we are doing, then not just sparrows, but other birds will also leave, seeking refuge in areas with cleaner air, food, and trees. Her words proved true. Over the years, I hadn't seen a sparrow, and even the number of crows has greatly reduced. Some bird species have also gone extinct due to pollution and other issues caused by humans. But then something unexpected happened.

THE IMPACT OF COVID LOCKDOWNS TO THE NATURE

During the COVID lockdowns, human activity was greatly reduced, and pollution decreased drastically. As a result, the birds that had left us due to pollution and other issues began to return to my neighbourhood. Though their numbers were still small compared to my childhood, it was clear they were making a comeback. Unfortunately, there were not enough trees to accommodate those birds. Once the lockdown ended, pollution began to increase again, and no wonder that sparrows and other birds vanished once again.

This experience connects with what Dr. Mary Trudeau said: "Get the river systems right, and the fish will do the rest," which is applicable to every field. We must make our surroundings clean, and then nature will take care of the rest. Similarly, as Dr. Santosh Padhi pointed out with the mosquito example, our instinct to eliminate a problem can sometimes led to unexpected issues. For instance, he shared an eye-opening example: when we use toxic chemicals like phenyl in drains to eradicate mosquitoes, we often fail to realize that these chemicals can return to us. They can mix with groundwater and eventually contaminate our wells, lakes and rivers, ultimately reaching us in the form of drinking water.

We often create things for our benefit, but if the environment suffers in the process, the consequences can be very dangerous for us. In the long run, it's as if we're chopping down the very tree branch, we're sitting on. Therefore, we should anticipate the impact our actions will have on nature and other species; we must also understand their perspectives.

ANIMAL BEHAVIOURS AS ENVIRONMENTAL INDICATORS

We often observe ants moving their families while carrying eggs, which may suggest that rain is coming. Some animals, like snakes and rats, leave their homes and seek safety days before an earthquake. Elephants, flamingos, and domesticated animals like dogs can sense vibrations caused by tsunamis. Similarly, other animals display various behaviours that indicate they are trying to protect themselves. If we could understand the signals that animals are conveying, we might be able to predict and prepare for natural disasters more effectively.

THE SILENT SUFFERING OF SPECIES

Over time, due to our actions, we have lost countless species, and we are on the verge of losing even more. A glaring example would be rhinos, along with numerous other endangered species. One might argue that we humans know what problems animals are suffering from. The problem with this mindset is that, while we have acknowledged the broader issues like climate change and pollution, some animals can still find a way to survive in pain, while others cannot - they just die and eventually become extinct. Therefore, the approach should not be generalized, but rather specialized to address the unique challenges faced by different species. We can only help them if we understand their perspectives. So, let's talk about one positive development in this very field that might address this issue with the help of Artificial Intelligence.

DECODING ANIMAL SIGNALS FOR A SUSTAINABLE FUTURE

Would you believe if I say that understanding animals and other species could be possible in the near future? Would you believe that we may have figured out a way to communicate with them? I know what you might be thinking: "What is this? Are you writing some kind of science fiction? Are you telling us the story of James Cameron's next Avatar movie?" Well, I'm not.

Scientists are currently developing AI systems that can analyse the sounds and signals animals use to communicate. To understand how this AI TO INTERPRET ANIMAL work, we need to first understand how AI works, so without going too technical, lets understand how artificial intelligence works.

HOW ARTIFICIAL INTELLIGENCE WORKS: A SIMPLE EXPLANATION

Modern AI is a complex field, but we can begin to understand it by exploring some basic concepts in computer science and how they have evolved.

THE FOUNDATION: IF-ELSE LOGIC

If you have read any programming language, like C, you might be familiar with conditional statements that is 'if-else' statements. But don't worry if you haven't or don't remember; I'll try to give you a simple example to help you understand.

This "if-else" statements serve as decision-making instructions for computers:

If a condition is true, then perform a specific action. Otherwise (else), perform a different action.

For example, using the if-else statement that we learned, let's create a SMART WATERING SYSTEM for your garden. This system will ensure that plants do not suffer when you forget to water them, ultimately benefiting our environment and supporting local wildlife, such as birds. By utilizing basic intelligence, the system will save water and automatically water the plants as needed. Please refer to the image below (IMG #1). I wrote a simple computer program in Python using just if and else statements.

IMG #1

# SMART WATERING SYSTEM V0.1
if rain_sensor.detect_wetness():
    print("The soil is wet, don't water the plants.")
else:
    print("The soil is dry, water the plants.")
# The above dummy code is for the ANIMAL INTELLIGENCE article: Shree

This code (Ref: IMG #1) uses a device called "rain_sensor” with the function "rain_sensor.detect_wetness()" to check if the plants need watering.

At this point You can see that our SMART WATERING SYSTEM is very limited and basic; it only provides water if the soil is not wet. If the soil is slightly wet, it will skip watering the trees. However, if it lightly rains, that might not be sufficient for the trees. Therefore, you can enhance it by adding additional intelligence to it. You can incorporate more 'if-else' statements to make it more capable. So, let's take a look at how we can do this.

IMG #2

# SMART WATERING SYSTEM V0.2
if rain_sensor.detect_wetness():
    if soil_moisture < 50: # If the soil moisture is below 50, the soil is half dry
        print("The soil is half dry, water the plants.")
    if 50 <= soil_moisture < 100: # Soil is between 50% and 100%
        print("The soil is okay, almost wet, check again later.")
    if soil_moisture == 100: # Soil is 100% wet
        print("The soil is full wet, don't water the plants.")
else: # No wetness detected
    print("The soil is 100% dry, water the plants.")
# The above dummy code is for the ANIMAL INTELLIGENCE article: Shree

Unlike the previous code (Ref: IMG #1), we used multiple if statements here to add more intelligence in controlling the device (Ref: IMG #2). A moisture level of 0% indicates dry soil, while 100% means fully wet soil. We set some conditions:

  • If the moisture level is less than 50%, we consider the soil to be half dry and water the plants.
  • If the moisture level is between 50% and 100%, we consider it okay, almost wet, so we instruct the system to check again later.
  • Lastly, if no moisture is detected at all (the 'else' condition), it will water the plants.

To understand the code a little better, here is some context:

  • “<” or “>” means "less/greater than".
  • “<=” or “>=” means "less/greater than or equal to".
  • “==” means "equal to".

Text following a '#' in Python is a comment for reference only and helps clarify the code.

Of course, to achieve this, you will need to do more programming and gather components such as a pump, microcontroller (an Arduino or Raspberry Pi), relay, and a way to determine whether it has rained (You can use a soil sensor and a weather API), along with some other components. This is just a rudimentary way to demonstrate how you can achieve some intelligence which is artificial using the concepts we just learned.

Traditionally, computer programs relied heavily on these if-else statements to make decisions. This type of decision-making mimics intelligence, earlier many software applications that market themselves as "AI" are often created using this if-else approach. Now, imagine a software program that can adapt its behaviour based on your actions or specific situations. This would require countless if-else statements written by the programmer to account for numerous scenarios, making it incredibly complex to maintain and difficult to build. This is where the modern AI comes into play, it goes beyond just if-else statements. Instead of manually coding rules for every possible situation, the modern AI systems learn patterns from data. It's as if they generate their own "if-else" statements based on the situation or data, but in a much more sophisticated and efficient manner than humans do.

HOW MODERN AI, LIKE CHATGPT, WORKS AND LEARNS

Unlike traditional programming that relies on fixed rules that we just discussed, Modern AI systems, such as Large Language Models (LLM) like ChatGPT, Claude and Image Generation Models like DALL-E or Midjourney, utilize advanced techniques inspired by the human brain, specifically "Neural Networks." It works like how humans recognize faces: by spotting patterns in features like eyes, noses, and mouths. They learn by repeatedly adjusting their understanding of data to improve pattern recognition and make predictions or decisions without explicit rules; this process is known as Machine Learning.

LLMs learn from data in a way similar to how a child learns by observing the world. This training process requires a significant amount of computing power, especially from large numbers of GPUs, which is why companies involved in GPU production, like Nvidia, have become one of the most valuable companies in the world, with a market cap surpassing $3 trillion a few months ago. Interestingly, with the right computer, you can even run and train an LLM offline.

TRAINING THE LLM

Data for training AI models comes from various sources, including books, articles, podcasts, publicly available social media content, and sometimes even private data that isn't fully public. Using certain data can lead to legal issues, so some companies create synthetic data to train their models instead. This approach helps mask or anonymize the training data sources, thereby avoiding legal troubles.

UNDERSTANDING LLM RESPONSES

After the complex training process, when you ask a question to the LLM through a chat window, the model uses what it has learned to provide a relevant response. However, whether it truly understands the output is a debatable topic. To better illustrate this, imagine a very smart parrot that has listened to thousands of conversations. When you talk to it, it can respond in a way that sounds natural because it has learned from all those conversations, even though it doesn't truly understand the meaning behind the words.

SIDE DISCUSSION: LIMITATIONS OF CURRENT AI TECHNOLOGIES

That's why many people do not consider large language models (LLMs) or any form of AI that exists today to be true AI; they are essentially sophisticated algorithms trained on vast amounts of data. These models process and generate text based on patterns in the data they were trained on, rather than exhibiting genuine intelligence or understanding. However, some argue that this is precisely what real intelligence is. They contend that the ability to analyse vast datasets, recognize patterns, and generate coherent and contextually relevant responses demonstrates a form of intelligence similar to that of humans. Moreover, LLMs can be biased. For instance, if a large language model were developed in Pakistan, it might generate responses that justify certain actions against India, depending on the narratives present in the data it learned from. This illustrates that these models are entirely dependent on the input data they receive. (I wish I could write this AI thing little more, but this is already out of the scope of this paper, so I stopped. Let me know what you think - you can email me your feedback).

DECODING ANIMAL COMMUNICATION WITH AI

Just as we train large AI models, such as ChatGPT, by feeding them vast amounts of human data, scientists are using similar techniques to decode animal communication. They input extensive datasets of animal sounds, behaviours, and other relevant information, then analyse these large datasets to identify patterns and meanings. Therefore, the key to decoding and being able to communicate with animals lies in understanding these patterns.

LEADING RESEARCH IN DECODING ANIMAL LANGUAGE

Projects like Project CETI (Cetacean Translation Initiative) are leading the way in understanding how animals communicate. It's a non-profit company, focuses on decoding the language of sperm whales by recording and analysing their clicks, known as codas, off the coast of Dominica. Researchers have even identified a "Sperm Whale Phonetic Alphabet," revealing how rhythm and tempo create diverse sounds. Another key initiative, BIOACOUSTIC AI, uses AI to monitor wildlife, studying the sounds made by different species to understand their behaviour, communication, and social interactions. This knowledge is crucial for conservation, as it helps identify threats, track populations, and develop strategies to protect endangered species.

By learning about the needs and challenges of various species, we can take more effective action to protect them. Recognizing animal voices can deepen our appreciation of their roles in ecosystems, inspiring more people to join conservation efforts and drive a collective movement toward sustainability.

HOW AI IS TRANSFORMING WILDLIFE AND ECOSYSTEM MONITORING IN INDIA

In India, AI-driven technologies are being engineered and helping us monitor animal populations and their habitats. By analysing data from various sources, such as satellite imagery and acoustic sensors, we are tracking changes in wildlife behaviour and habitat conditions. This information is important for making smart choices about land use and resource management. For example, AI cameras are now used to monitor forest fires in Similipal National Park in Odisha's Mayurbhanj district, helping us protect these important ecosystems.

THE BROADER IMPACT OF AI ON SUSTAINABILITY AND BEYOND

At this point, we know that AI is incredibly good at recognizing patterns, and with enough data, it has the potential to achieve amazing things. It can advance our understanding of the human body, aid in developing new drugs and vaccines, discovering new elements, and even help create efficient rockets and explore habitable planets. AI could also optimize solar energy use, assist in designing eco-friendly cities and infrastructure without negatively impacting nature, and ensure that other species do not suffer while combating climate change. Furthermore, AI could help us break the language barrier to communicate with non-human species. As we discussed Projects like Project CETI and BIOACOUSTIC are already using AI to decode animal communication, and in meteorology, AI models, like the one from Google, predicted 2023 Hurricane Lee's path faster than traditional methods. AI is also transforming healthcare, with tools that can detect early-stage cancer in medical images that doctors might miss. With AI, anyone can become a programmer, opening new possibilities for everyone. Additionally, advancements in AI can help address the challenges of current Quantum Computers and make them mainstream. This could lead to a new wave of innovation by combining AI with Quantum Computing. The future looks incredibly promising, what an exciting time to be alive!

NEGATIVE IMPACT OF AI ON NATURE

Every technology has its downsides, and the field of AI is no exception. While I won't delve into dystopian scenarios where AI rules over humanity, as depicted in science fiction, such discussions are for another day. The training and operation of AI models require substantial GPU resources, as we discussed earlier. This demand for GPUs translates into a massive consumption of electricity, much of which is still generated from non-renewable sources. As a result, the environmental footprint of AI can be quite large.

CONCLUSION: A CALL TO ACTION FOR A SUSTAINABLE FUTURE

While research on decoding animal communication is still in its early stages, the potential benefits are immense. By embracing these AI-driven technologies, we can enhance our understanding of the natural world and work toward a more sustainable future. The ability to communicate with animals could not only help us protect them but also ensure the health of our planet for generations to come.

In conclusion, as we pursue sustainability through engineering and AI technologies, we must acknowledge that AI's significant energy consumption and reliance on non-renewable resources present serious environmental challenges. By addressing these issues, we can ensure a sustainable future. let's remain open to the possibilities that lie ahead. The dream of understanding animal voices may one day become a reality, allowing us to collaborate with nature in ways we have never imagined. Together, we can create a harmonious relationship with the environment and take meaningful steps toward saving our world.