What can Artificial Intelligence do?
February 10, 2021
You might be surprised to find out that Artificial Intelligence (AI) is not that modern, as it has been established as a field of study since 1956 at a summer workshop held on the campus of Dartmouth College (US). Those who attended became the leaders of AI research (John McCarthy and 9 others). Before them, Alan Turing has conducted some research on AI, creating the Turing Test. The Dartmouth College gang failed by underestimating the difficulty of the project, bringing on the so called ‘AI winter’, when progress, interest and investment in the field stalled.
But at the beginning of the 21st century investment and interest in AI boomed. This was due to advancements in various fields that are tangent to AI. Innovations in hardware development, including sensors and GPUs, made it possible for AI to have the resources it needs to process huge amounts of data and to interact with its environment. Just like humans, AI needs experiences to learn from. Those experiences for AIs are the data sets, collections of data, whose availability has increased exponentially in later years. The development and refinement of machine learning algorithms and strategies have helped immensely, because these describe how the AI should process and extract information from the data sets or how to learn on its own. Also, Cloud services have made AI solutions available to more customers/businesses.
The victory of AlphaGo (a Go playing AI developed by DeepMind) against a professional human player, in 2016, is considered to be the starting point of The Age of AI. This victory was important because it shows that machines can really learn. The nature of the game of Go is such that you cannot use brute force to determine all possible moves, as there are as many as atoms in the universe. Therefore AlphaGo uses reinforcement learning and neural networks to mimic the learning process of a human brain.
Our world is currently going through a major transition. AI is progressing faster than you think and is very likely going to revolutionize the world, how we live and how we do things. Let’s see some of its applications in different fields:
Presumably, whoever masters the AI technology will rule the world, that’s why the first industry to adopt and deploy AI is the military and the greatest powers of the world are in on it (US, China, Russia).
Applications of AI in the military include:
- self-driving vehicles (airborne and land)
- lethal autonomous weapon systems
- intelligent surveillance
- robot soldiers
- robots that carry heavy loads over rugged terrain
- robots that can be sent into dangerous environments, like radioactive places, or for search and rescue
- anti-robot robots
- kamikaze drones
- earlier detection of terrorist attacks
If you want to be creeped out but also deeply impressed, check out the demo videos for the Atlas and Spot robots from the Boston Dynamics company. Their mobility, agility, speed and human-like movements are remarkable.
Some of the diseases that humanity has to deal with have too many variables. No human can hold in their head all the information necessary to understand them. But a machine can easily do that. Healthcare is a field where AI will feel at home and it can help make the shift from reactive to predictive care.
After studying enormous amounts of patient data and scans collected from all over the world, an AI can see patterns, it can diagnose, it can offer suggestions, predictions, it can detect shades in a scan that humans cannot and it can detect cancer 5 years prior to it being visible to humans. Among the number of patients that undergo surgery for potentially cancerous breast lesions, only some of them actually need these surgeries, and AI would be able to filter out the healthy patients.
Bionic limbs are already used with success and they use machine learning to make quicker decisions and improve. They improve the reading and interpretations of the signals from the muscles and also create the sensation that the artificial limb is part of the body by sending feedback to the wearer. Hugh Herr has become the pioneer of bionic legs after losing his legs due to frostbite. Who better to understand the physical and engineering challenges than an amputee biologist mechanical engineer?
Parkinson’s disease could be diagnosed earlier by wearing an AI device that detects changes in the voice or unusual patterns in walking and movements.
Accessibility devices are developed and available for blind people. These devices and phone applications use AI to interpret and describe the environment to the blind person.
Elon Musk’s Neuralink project is trying to resolve spine and brain problems (paralysis, brain damage, seizures, extreme pain, depression, restore eyesight, etc) with a seamlessly implanted device in the brain. The applications of this device are quirkier than that, going towards making new forms of communication between people possible, like telepathy. And offering a brain-machine interface that will allow us to control devices with commands sent directly from our brain and eventually achieve symbiosis with AI.
5G will enable more advanced AI. Remotely performed surgeries need precision and instantaneous timing that can only be delivered by the 1 millisecond latency of 5G.
Big agricultural machines can be driven and controlled from the living room of the operator. The machines are equipped with sensors and video cameras maximising safety. They can even remember where the seeds were planted, so nothing is left behind during harvest.
With computer vision, they can detect weeds, pests, soil defects and defective crops. The localized detection of weeds and pests allows for local and automated treatment, thus decreasing the quantity of pesticide used.
A farmer in Japan used AI computer vision and robotics to enable cucumbers to be automatically selected and categorized, saving him hundreds of work hours. The artificial intelligence driving his farm it’s called TensorFlow, it’s open source, and was created by Google.
Bicycles and trains were invented in the early 19th century, motor cars in the 1890s, aircrafts in 1903 and next on the list are electric and self driving or autonomous cars. Probably the first greatest and complete achievement in the field of AI will be that of self driving cars. This has been under serious development for years, with Tesla being well known. Smart highways have to be built to ensure the safety factor and gigafactories have to be built to cover the demand of lithium-ion batteries, so it will take some time.
The trucking industry, transportation of goods, is deemed to be the first that will be automated due to lack of personnel.
Speech and text processing
Humans can speak, listen and communicate through language. This is the AI field of speech recognition. Humans can write and read text in a language, this is the field of NLP or Natural Language Processing.
Advancements in speech recognition and NLP have brought us more and more accurate language translation tools and apps that transform text to speech and vice versa. This resulted, for example, in automated subtitles and bot narrated videos. Also many devices that are operated with voice commands have flooded the market: Alexa, Google Assistant, Siri, GoPro, curtains operated by voice commands.
There are text processing tools that can detect the tone of the text, like Grammarly or textio.com.
Grammarly plans to detect if the text sounds confident, accusatory, angry, sad, encouraging, egocentric, neutral, optimistic, regretful, excited, disapproving, worried, forceful, surprised, curious, rude, friendly or formal. But their AI is still in its learning phase, probably after a few years it will become more accurate.
textio.com is recommended for the hiring process, to help with formulating job requirements. For example, there’s a scarcity of females in the STEM fields, so to make this kind of job more appealing to women, textio gives recommendations on how to make the job requirements sound more feminine.
JPMorgan developed an AI that saved them 360,000 hours of man work each year. The machine does the mind-numbing job of interpreting commercial-loan agreements in seconds.
GPT-3, developed by OpenAI, is also making a buzz. Today, it’s one of the largest and most powerful machine learning models for natural language processing in the world. It generates tweets, pens poetry, summarizes emails, answers trivia questions, translates languages and even writes its own computer programs, all with very little prompting.
In the field of finance, AI has infiltrated the RegTech and stocks trading fields. It offers regulatory functions for broker-dealers, by detecting market manipulations like spoofing, layering activities, where a trader places multiple orders with the intent to cancel, to create the illusion of liquidity. AI powered trading bots are used to maximize profits and for micro trading.
Credit card fraud detection is another application of AI in finance.
Businesses and Marketing
Customer service for most businesses have been enhanced with various helping bots and chat bots.
AI can help with forecasting models, for example you can detect the demand for a particular color of clothes in a certain season/location. It can help with market surveillance, detecting patterns and anomalies.
An impressive application of AI is in personalized ads and suggestions that users see on various websites. Sometimes they are incredibly spot on, but behind that accuracy lies the study of your person, as you are being categorized and predictions about what is most likely to your taste are being made. This is done based on different variables like your searches, visited sites, your expressed interests online, age, gender and other. You can notice this on Facebook and Youtube, in the ads you see and in the pages that are suggested to you, on Netflix in the match percentage and suggestions. This is very helpful, it’s like having a mini personal assistant that reads your mind (but it can also pose privacy issues).
‘Just walk out’ is an AI powered technology that allows for stores without cashiers or lines. You install an app on your phone, walk in, take what you want, the AI powered supervision detects what you took, and when you walk out you will be automatically charged for the products and a receipt will be sent to your email.
‘NotCo’ uses AI to study molecular composition of animal based food and determine how to obtain the same taste, color, texture using plant based sources.
Lights-out factories are fully automated and need no human intervention. FANUC is a Japanese robotics company that has been operating as a lights-out factory since 2001. The robots in the factory can run unsupervised for as long as 30 days at a time. In the Netherlands, Philips uses lights-out manufacturing to produce electric razors, with 128 robots made by Adept Technology. There are only nine human quality assurance workers who oversee the end of the manufacturing process.
The applications of AI are much more numerous than in these fields. You will find AI in smart cities, in Tech Art – with robots that can write screenplays and can paint. In earthquake prediction, environment conservation and on Youtube video analysis that automatically removes suspicious content. AI can also recognize sounds (SoundNet) and it can generate images (Petalica-Paint, pix2pix).
An interesting job title is Head of AI for Social Good at Intel. The person holding this title is involved in AI powered anti-poaching devices. The devices are installed in the wilderness and take motion activated pictures. The pictures are then AI processed and if the presence of 4 human is detected the rangers are notified. This job is overwhelming for people given the quantity of photos and the need for immediate action, so an AI really helps with this.
In theory, automation and artificial intelligence should free us from dangerous, boring or repetitive tasks. It should help us be more productive and improve our overall quality of life. But there is a lot of fear around humans being superseded by machines and it’s not all unjustified. It is predicted that by 2030, 50% of the jobs will be lost to AI and humans are supposed to focus on more intellectual, creative jobs and on personal development.
The future can look dangerous and uncertain when advancements in AI are growing exponentially, but hopefully everyone’s active involvement in the subject matter and organizations like The Future of Life will be able to protect humanity’s best interest.
Do you have any questions?