A Cruelty-Free Future With Machine Learning

POSTED BY   Michele Baker
19th July 2018
A Cruelty-Free Future With Machine Learning

We know that artificial intelligence is being used in a vast number of ways to change the future. From alleviating us from the drudgery of boring work to providing medical diagnoses and even teaching our children, there are so many applications being put to work to make the world a better place. So, I wondered if the same technologies that are improving human lives could also offer a solution for helping other species break free from oppression, cruelty and suffering. After all, as a species we treat other animals appallingly, some might say primitively. Humanity has always sought to improve itself, to achieve ongoing enlightenment, and technology often plays a key role in this.

Animal testing in the pharmaceutical and medical industries is a necessary evil of our times. Whilst cosmetic testing has no justification, there is sadly no escaping that the only way to test new drugs with acceptable levels of accuracy is by torturing and often killing animals. Our diets, too, depend on shocking levels of savagery towards animals on a global scale. Animals are kept in captivity with unacceptable consideration of their health and wellbeing. Something needs to change.

Disclaimer: I write this piece in full knowledge of my own hypocrisy. I do, after all, eat meat (though never beef – I have a fondness for cattle), eggs and dairy products (yes, I know dairy hurts cows too). There is no justification for this, only excuses, as I’ll go into a bit later.


The Cruelty-Free Lab

Whilst it’s not quite perfect yet, AI News has this week reported on research taking place with a view to replacing animal subjects in lab tests with algorithms which will ascertain the viability certain combinations of chemicals for human use.

Using a statistical model combining both supervised and unsupervised learning, researchers utilise the K-nearest neighbour algorithm trained on random forest and logical regression algorithms to label new compounds. The AI is fed a database of 80,908 chemicals labelled according to their irritation level, corrosiveness and environmental impact. From this, the system is able to label new compounds and output their potential viability.

In animal testing, results are accurate up to 90% of the time, whilst the algorithm still falls short at a maximum of 80% accuracy. But whilst 57% of animal tests are fatal, no creature dies when using the algorithmic approach.

It’s early days, of course. The research is still ongoing, and would need to surpass the accuracy of animal testing substantially before being considered a practical alternative. Currently, ~£2.7bn is spent on animal testing annually. The use of AI could vastly reduce this, which could have a knock-on effect of reducing drug costs and thus saving lives (both animal and human).


Pain Relief For Sheep

In other news around the subject of artificial intelligence for animal welfare, an article from 2017 published in Newsweek looked at research taking place at the University of Cambridge. It seems that researchers at the university have developed a machine learning system that can estimate the level of pain being suffered by sheep. The system uses facial recognition technology trained on a database of 500 images of sheep to assess pain on what they call the ‘Sheep Pain Facial Expression Scale’.

Sheep Pain Detector AlgorithmIt’s been observed that there are five distinct things that happen to a sheep’s face when it experiences pain. Their cheeks tighten, their ears fold forwards, their lips pull down and back, their nostrils change into a V shape and their eyes narrow. Interestingly, according to Dr Marwa Mahmoud, co-author of the study, these facial actions are similar to those which occur when humans are in pain.

Whilst it is harder to ‘normalise’ a sheep’s face for a machine learning model than for a human (have you ever tried asking a sheep to pose for a photo?) the model still demonstrated 80% accuracy.

It is anticipated that the system, once perfected, could be used for early diagnosis of things like foot rot, allowing treatment and pain relief to be administered sooner and thus improve the animal’s welfare.

A similar system is also being used at Maruyama Zoo in Japan in order to monitor the health and wellbeing of the animals in its care. Those working with the system also hope to extend its use to dairy farmers and hospitals.


Tastes Like Chicken

There is also a light at the end of the tunnel for those who recognise the damning ethical and environmental effects of the meat industry. Currently, meat production accounts for 60% of global greenhouse emissions, up to 90% of rainforest destruction, and takes up a disproportionate amount of resources like water, crops and land. Veganism, like it or not, is the only ethical solution and the only way to create a sustainable future for humanity.

Vegan Plant-Based FoodHowever, most of us are reluctant to give up our current diets, as damaging as they are to our bodies, animals and the environment. Switching to a plant-based diet means foregoing the tastes to which we have become accustomed, as well as the convenience of sticking with mainstream culture. I know from experience that when I trialled a vegan diet myself, it was all but impossible to pick up a simple sandwich for my lunch that didn’t contain at least one animal product. Meat and dairy are so ingrained in our lives, it can feel like it’s impossible to quit. That’s not to mention the teasing, horror and sometimes downright disdain you get from friends and family for daring to be less than convenient at mealtimes.

Most people are aware that technology is being used to try and create lab-grown meat that is cruelty-free whilst tasting identical to its ‘real’ counterpart. I welcome this work, and am eagerly awaiting its arrival on the shelves. Not everyone is so keen, being so devoted to the status quo and the narrative of the meat industry that tells us we downright need to kill and eat animals to get the nutrients we need. Untrue, by the way.

Founder and CEO of Hampton Creek, Josh Tetrick, has a passion for pushing the world towards a sustainable, cruelty-free, plant-based diet. His company uses a machine learning algorithm to systematically find new ingredients and formulations from all the world’s edible plants that could provide substitutes for animal-based products. The AI uses reinforcement learning to improve its suggestions, and has already come up with some viable alternative foods. It identified an isolated protein in the mung bean that has similar properties to scrambled egg, and has also suggested a rather impressive new chocolate prototype that uses goji berries, mushrooms, broccoli and an undisclosed nut. Sounds awful, but the entire system is designed to completely mimic the taste of original foods, so chances are, diners will be impressed.

Mayonnaise, milk and cookie dough have also been created using the system, but solid foods are proving a little more challenging. Research continues, however, and already the results are promising. The foods taste identical to the originals, but have the same or better nutritional profile.

Simply knowing that scientific minds are working hard to create a sustainable, cruelty-free future goes a long way towards providing hope that our children will live in a better world than our own. As we continue to find new and improved ways of feeding ourselves, of providing vital medicines and caring for creatures in captivity, we grow as an enlightened species. But old habits die hard, and it will take nothing short of an entire change in mindset to implement these promising new technologies in the real world. Despite the challenges, we must continue to try. So very much depends on it.

Don’t worry, we’ll never share details with anyone.

A Cruelty-Free Future With Machine Learning

Michele Baker

Michele Baker is the Senior Content Strategist at TDMB. She began her journey into tech marketing via a Masters in Creative Writing, evolving from a prize-winning poet and short story writer to a futuristic content guru. Michele now writes endlessly about all aspects of technology, hosts the TDMB Presents… tech podcast, and speaks at numerous tech and marketing events.

Get in Touch With Michele Baker

01306 632 854

You may also like

Machine Learning In Retail – A Help Not A Hindrance? 
17th January 2018
Machine Learning
Machine Learning: The difference between knowledge and understanding
The Sunday AI Review | 18-24th September 2017