Natural sciences 🌱
Bioinformatics, geology, pharmaceutics, automotive industry, astronomy… There are countless natural science fields using robotics and AI. The use is common = to discover. Nowadays, robots are the companions of human scientists and explorers. Even though AI development still has a long way to go, it is already advanced enough to replace some operations that humans could never do so accurately and efficiently.
One of the most useful AI skills is computer simulation. A pre-programmed algorithm with large amounts of data can predict and simulate a given experiment very accurately. It is also a safe way of research, since humans are not so good at predicting and they often bet on fate, which can cost them the most precious thing = life. Take chemistry as an example. Chemistry is a subject that is now taught in high school and those who are interested continue to study specialized areas of this subject. Chemistry as an scientific field has existed since the 17th century, but it only began to grow dramatically from the 19th century to today.
The occasion of discovering something new
In recent years, AI has begun to help in many areas of chemistry (e.g., organic and inorganic chemistry, biochemistry), but has been most prominent in the field of analytical chemistry. Analytical chemistry deals with the study of the chemical composition of substances and mixtures. Artificial intelligence can thus be practically applied to a variety of tasks where complex relationships are often present in data sets. For example, AI can predict the solubility of a new compound using either equations based on empirical data or theoretical calculations. However, solubility predictions can also be made by an AI program that has developed structure-solubility relationships after being “trained” on many compounds with known solubilities. The use of AI for tasks such as property prediction has expanded in recent years due to rapid growth in computing power, more open-source machine learning frameworks, and increasing data literacy among chemists.
Criminalistics 🔦
Criminalistics is a broad field that includes, for example: analysis of DNA, fingerprints, bloodstains, firearms, or analysis of digital audio video tracks and photographs, and much more… AI can be very well applied to profiling people and “predicting” their actions.
Safer society with AI?
Through video and image analysis, AI improves law enforcement outcomes by reducing time-consuming tasks and human errors. One of the key skills of AI is facial recognition. They can determine the identity and whereabouts of an individual, which greatly improves crowd surveillance results.
In facial recognition, AI assesses clothing, skeletal structure and body movements to detect suspicious citizens in society, such as shoplifters or dangerous drivers violating traffic laws. It also helps with vehicle identification, as AI programs learn to decipher license plates even in poor resolution or low ambient light. Several governments have already authorized the use of AI in law enforcement, such as the Canadian police.
Artificial intelligence can be very useful in detecting traffic accidents through closed-circuit television (CCTV) monitoring with installed security cameras and in detecting online crimes including human trafficking, money laundering, fraud and sexual abuse. By detecting suspicious activity, AI can prevent crimes and help investigators identify suspects more quickly, ensuring greater public safety and increased community confidence in human law enforcement and criminal justice. [1]
Medicine 🧬
Robotics and artificial intelligence have huge potential for healthcare. As technology evolves, robots work more autonomously and they’re likely to perform certain tasks entirely on their own in the future. This will allow doctors, nurses and other healthcare professionals to spend more time providing direct care to patients.
Surgical robots
In practice, robots are already commonly used in surgery. There is also the field of so-called surgical robotics, which is still evolving. High-resolution 3D computer vision allows surgical robots to distinguish different types of tissue and provides human surgeons with detailed information and enhanced performance during surgery. For example, surgical robots now have the ability to help surgeons avoid nerves and muscles during procedures. The vision for the field is the ability of robots to take over small sub-procedures, such as suturing or other pre-defined tasks, under the watch of a professional human surgeon. This allows procedures to be performed more efficiently and accurately. Did you know that company Intel technologies provides the foundation for robotics in healthcare?
“Intel is working with technology providers and researchers to explore the next generation of robotic solutions. By providing technology and research support, Intel is helping to explore new applications for AI and IoT technology in medical robotics. These contributions support ongoing innovations that increase automation, efficiency and solve some of the biggest challenges in healthcare.” [2]
Note: IoT stands for Internet of Things, which is Intel’s technology that helps drive (not only) medical robotic machines thanks to their powerful processors.
Google’s algorithm saves lives
AI algorithms are great for automating difficult tasks and can sometimes outperform humans at tasks they are trained to do. As an example, an algorithm coming from researchers at Google AI Healthcare created a learning algorithm called LYNA (Lymph Node Assistant) that analyzed stained tissue samples on histology slides to identify breast cancer tumors from lymph nodes. Interestingly, such an algorithm could in the future identify suspicious areas that are indistinguishable to the human eye. LYNA was tested on two datasets and was shown to accurately classify a sample as cancerous or non-cancerous 99% of the time. This research was conducted in autumn 2018. [3]
Army ⚔️
Utilization of AI in military forces is one of the most controversial topics related to AI. Autonomous weapons, software predicting the course of a potential war conflict and unmanned tanks or drones are commonly used in modern tech-advanced militaries. AI powered weapons are able not only to predict and evaluate the situation, but also to detect an enemy. Nowadays, soldiers have these weapons under their control but there is an apprehension… We’ll get to it later in this article.
A brief look into history
Although the popularity and some kind of awareness of AI came quite recently, if we look at the history of AI used in military, we need to look back to the Second World War. A British machine called “Ultra” had been developed to break high-level encrypted communication of German Enigma. Thanks to Ultra, the Allies could get to know what will be the next move of the enemy and according to that information they could discuss and proceed on next, more effective move. This moment officially doesn’t count as the very first usage of AI. They didn’t even know at that time (also the term “AI” itself didn’t exist), but nowadays we might consider it because the way the machine worked could be described as a simple prototype of AI.
Role of AI in military
Nowadays, the role of AI in military is clearly defined:
- To erase errors that are coming out from human inaccuracy
- To boost effectivity of military operations
- To perform more precise attacks with as few losses as possible
To look at what exactly AI can do in military, let’s look at some examples. First, the Iron Dome – the air defense system used in Israel to predict the trajectory of enemy rockets and to decide whether it should be destroyed. Iron Dome will shoot down a rocket if it detects that the rocket is directing to an inhabited area. Second example could be analyzing data. People don’t have that much ability to process big amount of information and we naturally make mistakes. On the other hand, AI is capable of processing tons of data and find mathematical formulas at the same time. In an enormous speed.
Thanks to these features, AI can predict a situation or needed supplies. It can be used to create simulations for training soldiers, to explore new area and simply to make military more effective and prepared for potential dangers.
A strong teammate, but an even more dangerous enemy
Now we are getting to the problem, that probably no one knows how to deal with it. The simplicity of manipulation of AI is the warning sign of how dangerous it can be. You’ve probably heard about deep-fakes. These are fake images or videos replacing faces and speech of influential people (e.g. politicians) at so accurate level, that most people actually believe it. Following manipulation, disinformation and fear in the public may lead to an information war. Also these advanced weapons mentioned above owns only a few wealthy countries in the world. Of course the rest want them too so they could maintain the pace. However, that would only increase the chance of the next world war, especially if those weapons would get under the hands of criminal organizations. But above all those dangers, still the biggest fear is that humans might completely lose the control of AI.
How could robots and AI theoretically take control of us?
First, there is a possibility that robots would create a completely new language that people wouldn’t understand. This actually happened already in 2017, when despite two Facebook chatbots were programmed to communicate in English, they invented an own language anyway. Because of fear that humans couldn’t understand what they were talking about, they were immediately shut down. Imagine if robots created a new language and started to decide and take actions without human supervision or reflection, it would mean a catastrophic scenario for us. With machine learning, they would be able to learn way faster than us. In the past decades, the biggest threat was to start a nuclear war. In the present, the biggest fear is that robots would launch nuclear warheads without any human intervention…
I believe, that the only way to prevent getting AI out of our control is to regulate it and to achieve some kind of international cooperation for controlling and limiting the usage of it.
Art and humanities 🎭
AI is also currently a great help in fields more focused on nature and humanity. But since these features are “unpredictable”, it is much harder for a robot to understand natural language, for example, than to calculate a complex formula. This is due to the fact that there are no formulas or boundaries for language. In mathematics, everything has a fixed logic, this may not be the case with language or art. From this point of view, the humanities are flexible, which seems to be a big obstacle for robots. Yet there are machines and programs with at least limited creative thinking.
DALL-E: The most creative AI software
One of the most popular AI programs that is capable of creating artistic images is DALL-E. This program can generate images based on any description in natural language (English). It can also combine different ideas, which is one of the key elements of human creativity. In September 2022, a second version was released: DALL-E 2 for anyone in the world and it is free to a limited extent. Images of what this program can create immediately started appearing on the internet. DALL-E 2 is even able to create different options and combinations and then user chooses what they like the most or, on the contrary, can change the text they have entered. But of course it has its flaws, because it only understands natural language to a limited extent.
Sometimes he is not able to distinguish the difference between “Yellow book and red vase” and “Red book and yellow vase”. Also, the requirement for more than 3 objects, negation, numbers and joined sentences can result in errors. What is amazing, however, is that this program was used to generate the cover image in Cosmopolitan magazine, as a world’s first. The request was: “a wide-angle shot from below of a female astronaut of athletic build swinging a camera on the surface of Mars in infinite space.” [4] The image was not further manually edited and took 20 seconds to create.
Open-AI strives for achieving creative thinking ability
DALL-E is being developed by Open-AI and works on the principle of deep learning. However, based on security, which became an issue immediately after the program was launched to the public, certain results had to be regulated, especially various cues.
In December 2022, a sibling was released: POINT-E, which is basically the same but generates 3D images. POINT-E is far from perfect and doesn’t yet work as easily, i.e. by simply entering text, as DALL-E, but it’s a great start for modelling 3D graphics.
DALL-E isn’t the only program that Open-AI has created. You’ve certainly heard about ChatGPT. On March 2023, the forth version was released to the public. ChatGPT is an advanced language model that can work with text input very well. GPT-4 is more creative and collaborative than the previous versions of this model. It can generate or edit creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style. And English isn’t the only language that this model understands!
It is more difficult for a robot to understand language than any complex mathematics
As for languages, that’s where it gets tricky. Google’s translator, for example, is far from perfect, but it has improved over the last few years and its translations are more accurate. In 2016, Google introduced the GNMT (Google Neural Machine Translation) system, which uses an artificial neural network to improve the fluidity and accuracy of translations. Human languages are particularly complex, and linguists thought for a long time that it would be impossible to teach a machine to analyze speech sounds and word structures the way humans do.
However, researchers from MIT, Cornell and McGill Universities have created a new machine learning model that discovers linguistic rules. These often coincide with rules created by human experts. This machine learning model has learned that the letter “a” must be added to the end of a word to make the masculine form feminine in Serbo-Croatian. For example, the masculine form of the word “bogat” becomes the feminine “bogata”. [5]
Education 🎓
The world and society are constantly changing and we already know this for a long time thanks to history. However, changes have taken place at an unprecedented speed over the last century, especially when it comes to the development of IT. As a person from Generation Z, I find myself in a unique position. This generation is actually the first one in the world that grew up at a time when current technology and the internet already existed and started to be an everyday thing.
We are the first generation to experience problems with mobile and social media addiction, and we are seeing an increase in social anxiety, as we often prefer to communicate via chat from being in touch with people, including our loved ones. We also have access to all kinds of information, and are often overwhelmed by it.
And this trend will continue apace. Gen Z is no longer the youngest generation. New generations are coming and will come and they’ll be even more accustomed to current and future technologies. Already from this point of view, it must be taken into account that many things will need to be changed or upgraded. What was true yesterday may not be up to date today. And this also (and above all) applies to the way in which current and future children and students are educated.
Adapting education for future generations
The modernisation of education is necessary and inevitable in the current world, although some of the older generations do not want to admit it. Mostly they do not want to change educational practices and methods not because they think it is unnecessary, but because they simply don’t understand it. Whether they want to understand it or not, that an another story.
Today’s technology offers new and innovative solutions for learning in any subject. It is clear that, for example, office suites (Office 365, G-Suite, etc.), email or generally modern IT equipment should already be standard in all schools. Yet not everywhere this is the case today. So my question is, if we want to implement more technology in schools, including artificial intelligence and robotics, why doesn’t every school have even the most basic IT equipment in the first place?
Modern equipment and robotics integration for future-ready students
As mentioned, a computer lab with modern desktops and laptops should be standard. If pupils do not have quality equipment at home, they should at least have it at school. What does modern equipment mean? Simply that one that is sufficient for school work. This includes, of course, graphic design or programming. Pupils should not be limited by the performance limits of that computer.
Robotics could become an entirely new subject. Building and programming robots could be fun for some students and possibly a new interest. It would also prepare them for their future if they want to continue to pursue robotics. So why not give them that opportunity? Today, there are programmable robots or kits such as Arduino or Lego Mindstorms, which are more suitable for younger children.
There is also an interesting question about the use of VR (virtual reality). But why should we introduce VR in schools? Wouldn’t that be “too much”? Wouldn’t it be too unnecessary? About VR in education, visit my another post focusing more on this topic.
More effective teaching through personalised learning
Today, thanks to various apps, learning is also faster and more personalised for each of us. Artificial intelligence will help supplement the work of the teacher in the classroom and identify some weaknesses in the classroom that the teacher does not immediately see. For example, artificial intelligence could recognize when some students are missing specific questions. By alerting the teacher, they will know that they need to relearn the material because the students don’t yet understand it. Teachers will then be more in the loop and follow best teaching practices.
Traditionally, students only get answers to their questions when they meet their teachers or professors and have the opportunity to ask them in class. Fortunately, AI can replace it. There are special AI-powered chatbots created for education. They work as assistants to students to answer their questions whenever they want. So, students don’t have to wait until meeting a professional in their office or classroom.
“Anyone who’s informed about global trends across different industries will know that personalization is top of the list. This is due to the advent of artificial intelligence, which is an advantage for the education sector. AI helps teachers up their game, providing them all the information that they need. It also allows teachers to create content that suits their students best while ensuring personalized learning. It automates tasks, so teachers have more time to do more teaching and impact the students better.” [6]
Is implementing AI to education even possible?
As mentioned before, AI offers a wide range of possibilities in education. But the big question is how to apply these gadgets to real schools. First of all, such technologies need to be developed first. I don’t mean that the software options listed don’t exist. Such technologies already exist, but they are not well enough developed to be used every day in schools. Moreover, they are not so widely available.
Therefore, we need to invest funding and time in developing AI for education so that in the future AI programs can be accessible, safe and easy to use for schools. Furthermore, teachers would need to be trained. Teachers would have to learn how to work with such programs – be instructed and use them properly. Therefore, in the future, there will also be a need for people to manage AI systems, just as there is a need for people to manage servers in different departments and schools nowadays.
Concerns about privacy and teachers losing their jobs
One prevailing concern surrounding the integration of AI in education is the fear that it will replace teachers, leading to job losses. This is not entirely true. It’s crucial to understand that AI’s role in education is not to replace educators, but rather to enhance and support their work. In fact, in education, AI isn’t meant to replace teachers. It is meant to complement them.
Another critical point is security, especially when it comes to processing student data. AI tools are evolving to create more immersive learning experiences in which students use their critical thinking skills to access information. However, such tools come with drawbacks – and a number of potential risks. Machine learning and AI are at the heart of many data privacy issues today. By embracing AI as a complement to teachers and implementing robust data protection measures, we can unlock its transformative benefits while ensuring the safety and privacy of students’ sensitive information.
Resources:
[1] IDDER, Asma a COULAUX, Stephane. Artificial intelligence in criminal justice: invasion or revolution? [online]. 13.12.2021. Available from: https://www.ibanet.org/dec-21-ai-criminal-justice
[2] Intelligent Decisions with Intel Internet of Things (IoT) [online]. Available from: https://www.intel.com/content/www/us/en/internet-of-things/overview.html
[3] GREENFIELD, Daniel. Artificial Intelligence in Medicine: Applications, implications, and limitations [online]. 19.6.2019. Available from: https://sitn.hms.harvard.edu/flash/2019/artificial-intelligence-in-medicine-applications-implications-and-limitations/
[4] The World’s Smartest Artificial Intelligence Just Made Its First Magazine Cover. Cosmopolitan [online]. 2022-06-15. Available from: https://www.cosmopolitan.com/lifestyle/a40314356/dall-e-2-artificial-intelligence-cover/
[5] ZEWE, Adam. AI that can learn the patterns of human language [online]. 30.8.2022. Available from: https://news.mit.edu/2022/ai-learn-patterns-language-0830
[6] HARPER, Tiffany. Top 7 Ways Artificial Intelligence Is Used in Education [online]. 25.8.2021. Available from: https://trainingmag.com/top-7-ways-artificial-intelligence-is-used-in-education/