Artificial Intelligence: What is it Exactly?
AWiB is giving HER focus to a topic that is stirring up commotion around the world. Technical people want to be experts in it and most non-technical people feel as though it is an abstract idea. AWiB took it upon HERSELF to de-mystify this concept and get to the granular details of what it is, how it is used, what its potential dangers are, and more. Our “Focus” for this month is Artificial Intelligence.
In the beginning, there was philosophy.
The concept of innate objects evolving into an intelligent being is not exactly new. For instance, the Greeks had a mythology that describes a robot that protected Europa form pirates and invaders. First mentioned by Hesiod in 700 B.C, the story of Talos describes a tale of a giant bronze man built by Hephaestus, the Greek god of invention and blacksmiths. In his other work, Hesiod portrays Pandora, known by many for unwittingly opening the box of evil, as an artificial, evil woman built by Hephaestus and sent to Earth on the orders of Zeus to punish humans for discovering fire. It was as if Pandora was an AI agent that we are all too familiar with in sci-fi movies like The Terminator.
But while myths and stories existed for a long time, the field of AI was formally founded in 1956 at a conference in Dartmouth College. During this conference the term artificial intelligence was coined to the topic. Event attendees were extremely optimistic about how fast they would be able to create intelligent beings. But they failed to realize how difficult a task it is. After several reports criticized the progress of AI, government funding and interest in the field dropped. From 1974 to 1980, in what is commonly termed “the AI winter,” things were looking bad for AI. In the 1980s, the British government started funding it again, partially in an effort to compete with the Japanese. But not long after, the field saw a decline in funding from 1987 till 1993, with the collapse of the market for general purpose computers. Significant change was seen when—in 1997—IBM’s Deep Blue became the first computer to beat the chess champion, grandmaster Garry Kasparov. Then in 2011, the computer Watson won the quiz show Jeopardy by beating champions Brad Rutter and Ken Jennings. After such victorious runs, AI became popular yet again.
What does AI stand for?
For most people, the phrase “Artificial Intelligence” brings up a picture of a robot that is here to destroy the world; a robot capable of thinking, acting, and existing independent of human intervention. But before jumping to conclusions and immersing ourselves in fear, let us try to understand AI. Let’s start by dismantling the term itself: Artificial and Intelligence. Artificial means something that is made by humans. Intelligence refers to the ability to acquire and apply knowledge and skills. When you put those two words together, you get the ability to acquire knowledge by something that is human-made. But does the ability to acquire knowledge qualify as being equal or greater than a human being?
Scientists have dwelled over the idea of computers qualifying for human-level intelligence and more. This raises the question of the ability of a machine to trick people into thinking it is in fact a human being. To this end, famous mathematician Alan Turning came up with a test to determine if you are talking to a machine or a human. In the Turning test, a computer would be asked a series of questions by judges and at the end of the questioning, the judges decide if they talked to a human or a computer. A talking computer chatbot called Eugene Goostman got significant attention when it was able to trick one third of the judges that it was human. The bot was able to dodge some questions by claiming it was an adolescent who spoke English as a second language. Now scientists are doubting the solidity of the Turning test. However, passing or failing the Turning test does not change the fact that the machines are computers.
How does AI work?
So if AI is a computer, how is it different from a laptop or a PC at your desk? Computers are machines that we use to make our lives easier. It could be a PC, a smartphone or a robot with a remote control; all those devices are machines that we use to do things that require extra efforts from us. But for those machines to serve their purposes, we would have to tell them what to do. We tell them to open an application, or press buttons to order the remote controlled robot to go left. But with AI, you don’t need to explicitly tell it what to do. AI machines have the ability to learn things and act accordingly in the future, just like humans.
Take this example: You want to make a self-driving car. If it is a simple computer, at every step you would have to tell it what to do, where to go, when to stop, etc. Therefore, it defeats its purpose; it is not driving itself. If you use AI, you train it—like you would train a human—on a set of data and after it completes its training it can drive just like any human.
How is it being used?
AI is not necessarily a robot in its physical form with arms and legs. It could be as simple as an application on your phone or a chatbot on a website. Today, AI has made its way into many sectors.
One thing about humans is that we love to know anything and everything about the future. We want to know what will happen in a week, a month, a year or even a decade from now. For a long time, predictions and forecasting have been done using manual calculations and common sense. Farmers use traditional prediction methods whereby they utilize natural indicators to forecast the weather, drought, flood and the like. But while traditional forecasting was sufficient in the past, the constantly changing climate and the need for accuracy make the method outdated.
As for statistical predictions, where manual calculations are completed on a set of parameters to predict certain events, their lack of flexibility and adaptability to new types of information poses a drawback. AI, on the other hand, has the ability to adapt and change in accordance with new trends. It can acclimate to changes by training using data as it comes. Machine learning predictions have greatly simplified forecasting related to agriculture.
Medical science has advanced tremendously over the past few years. Nanotechnology has made it possible for scientists to use tiny robots that maneuver through the human body to detect and target specific cells. In this critical sector, AI has proved its invaluable use. Companies around the world are using AI for making more accurate diagnosis, significantly reducing the death caused by misdiagnosis. Additionally, AI is being used during medical research: assisting in development of drugs, choosing eligible clinical trial participants, and automating routine tasks like background checks.
With the rise of the internet, privacy and online security have become critical issues. People are constantly exposed to cyberattacks and a great number of people have lost money, properties and even their identities. Fraud detection has been and continues to be a subject of interest for intelligence agencies, private companies and governments. While there have been several methods that relied on set patterns and analyzed those restricted set of patterns, those methods were susceptible to failure when an attack did not fit into their predefined design. AI helps gain a better understanding of the customers’ behaviors, making it a good option for effective detection of new and emerging fraud attacks. AI-based technologies can be applied to payment-fraud detection, banking fraud, financial fraud and tax evasion prevention.
Autonomous flying machines have applications ranging from security surveillance to medical supply delivery; hence, improving how they operate is a primary concern for many stakeholders. One of the key areas of improvement is in computer vision and object detection. If a military drone misses a target, the results will be catastrophic. That is where AI comes in. Deep machine learning algorithms can be applied to enhance the drones’ object tracking, self-navigation, and obstacle detection mechanisms.
As for self-driving cars, the concept has gained attention after the interest of companies like Tesla and Waymo. Those companies have moved on into testing phase of the technology with hopes to fully automate the task of driving. What is interesting about automation of vehicles is that it comes in different stages:
- Driving assistant: the system does not take full control but rather assists—for instance in the case of parking.
- Partially automated: the system has partial control but the driver is responsible for the operation of the vehicle.
- Highly automated: the system can take control for a long duration such as on the highway.
- Fully automated: human presence is needed but the system takes full control.
- Completely automated: the vehicle can navigate without any assistance from the driver.
Most hotels and resorts depend heavily on their customer service to build their reputation. AI technology can assist with this in a number of ways. For instance, AI can be used to improve personalization, tailor recommendations, and guarantee fast response times regardless of the presence of staff. Hotels and other tourism-based businesses are extensively applying chatbots and online customer service. Some hotels are taking it a step further: Hilton Hotel uses the AI robot “Connie” to do face-to-face customer service. Connie uses AI and speech recognition to provide tourism information to customers who talk to it. AI has also proved its assistance in data processing and analysis for the tourism sector.
- Smart Homes
Ever wish that your fridge would automatically buy all the groceries and stock itself up? Well, the way science is heading, such fantasies might not be as far-fetched as you would think. Smart homes, which are homes that allow inhabitants to remotely control them using connected devices, are increasingly becoming popular. In a smart home, your washing machine, TV, refrigerator, and the like, are all connected over the Internet. Their connection is then tied up with an Artificially Intelligent system that will do the thinking and deciding. For example, Google’s Nest thermostat is an electronic, self-learning, Wi-Fi enabled thermostat that optimizes the heating and cooling of homes to conserve energy. Users will calibrate the device for the first two weeks during which the thermostat will collect data points and learn the schedules of the people. Then, it will automatically adjust the temperature and even go into energy-saving mode when no one is at home.
- Entertainment and media
AI has made its way into the media sector with applications in marketing and advertisement, personalization of user experience, and search optimization. Companies are training machine learning algorithms to help develop film trailers and to design advertisements. For example, 20th Century Fox used IBM’s Watson AI to develop the trailer for the horror movie Morgan. The AI was trained on scenes from 100 horror movies.
The AI then analyzed the scenes based on visual, audio and composition elements. After the system had an understanding of the type of scenes in horror movies, it was given the full length film and it recommended 10 scenes from the movie. Watson was able to make the trailer in 24 hours, a process that would normally take two weeks to complete. In addition to making trailers in unbelievably short time, AI is being used by companies like Netflix to manage workflow. Netflix announced the development of a workflow managing AI application called Meson to manage the company’s various machine learning processes that build, train, and validate personalization algorithms responsible for recommendations.
Who is using the most out of AI?
Although AI application and research is going on around the world, some countries have advanced more than others. In particular, China and the United States of America are on an ongoing battle to get the upper hand in the sector. China has held a strong ambition to become the AI Superpower of the world. The country has declared to be a $150 billion AI global leader by 2030. China has published a number of research papers on deep learning as compared to other leading countries. China takes advantage of the fact that its vast population uses Internet, generating a large supply of digital data for processing.
China’s nemesis in the race for becoming the world leader of AI is the U.S. Thanks to its well-established tech culture, the U.S. has benefited from $10 billion venture capital channeled towards AI. Many of the big tech companies like Google, Facebook, Tesla and Amazon are investing a lot of money into research on AI and its applications. Additionally, start-ups that work on AI have sprung up in the U.S. with Argo AI, a self-driving technology platform company leading the way with a funding of $3.6 billion.
Other countries involved in the game are the UK, Canada, Russia, Germany, India and Norway. The UK is the leader of the sector in Europe, with 121 AI-empowered firms. The UK government has also shown interest in the matter, with a funding announcement of $78 million to support AI and robotics research in 2017. Similarly, the government of Canada plays an important role in investing into AI projects with the fear of being left behind by China and the UK.
Where is Africa?
There is no denying how useful AI will be for developing countries. With applications in healthcare and agriculture, the technology will definitely bring forth significant change. To assess the readiness of AI in Africa, A Roadmap for AI for Development in Africa was organized in Nairobi with 60 African and international experts gathered to discuss how to ensure inclusiveness and best utilization. The roadmap includes the need for policy reforms and regulations, increased number of research centers and groups, funding support, and participation encouragement. It calls for the use of existing initiatives like Black in AI and ALLIANCE FOR AI to build capacity and strengthen training and research.
As for start-ups in Africa that work on AI, the market seems to be getting up to speed. Launched in 2014, Aerobotics is a start-up based in Cape Town, South Africa, with the aim of using machine learning to analyze maps and extract actionable information for crops such as wheat, citrus, and sugar cane. Another AI start-up, Tuteria, based in Nigeria is an edtech company that matches qualified tutors with students according to their area and budget. In Egypt, a start-up called Affectiva provides emotion recognition software solutions that use the company’s database to detect moods and make decisions based on facial expressions.
Where is Ethiopia?
Even with its significant share of challenges to advance technology, Ethiopia continues to be aspirational in getting involved in AI research and development. The AI research laboratory, iCog Labs was part of the team that developed the software for Sophia, the world’s first humanoid robot. With clients in U.S., Canada, Hong Kong and China, iCog works on many projects that have the potential to transform the dynamics of AI. In addition to the projects, the company has an initiative called Anyone Can Code, by which they teach high school and elementary students programming languages and machine learning technologies with the aim of building the next generation of technically enabled youth. Another AI based start-up in the country is Lesan AI, a machine learning based translation service that hopes to break the language barrier that is preventing many Ethiopians from accessing useful content through the Internet. Lesan translates online content to native Ethiopian languages with the Amharic translation already deployed and the Tigrigna service currently in testing phase.
On the government level, Addis Ababa University hosted the first ever Artificial Intelligence Conference (AIC) in Ethiopia from Oct 28-Oct 30, 2019. The conference was organized to coordinate and consolidate all Ethiopian researchers and scientists working the areas of AI in the fields of agriculture, health, and education. Additionally, Prime Minister Dr. Abiy Ahmed inaugurated the AI research and development facility on September 20, 2020. The establishment of the center was approved on January 24 of the same year by the prime minister’s cabinet. Dr. Abiy has vocalized his interest in the AI sector on many platforms, calling for the need to include the youth in the development of AI technologies. The AI research and development facility will serve as a support for young entrepreneurs through capacity building and research.
Women in AI
Globally, there have been active calls for the inclusion and participation of women in the field of AI. Organizations such as Women in AI are actively working to break the gender biased barriers and include more women in the field. As of 2018, only 22% of the AI professionals worldwide were women. The bias expands to affect women disproportionately when it comes to reaping the benefits of AI. Machines that have been trained by biased datasets go on to perpetrate that bias even further. To shed light on some of the women who are spearheading research in the field, here are some notable women working on AI:
- Layla El Asri: Research team lead at Borealis AI in Montreal
Layla focuses on research in Natural Language processing, Deep Learning and Unsupervised Learning. She completed her PhD in Philosophy and Data Science at the University de Lorraine in 2016.
2. Yejin Choi: Associate professor at the University of Washington
Yejin completed her PhD in Computer Science from Cornell University. In 2019, Yejin presented at 13 conferences discussing her latest work which includes her primary research interest areas of Natural Language Processing and Machine Learning.
3. Noor Shaker: CEO/ Founder Phenogeneca
With over 15 years of experience in AI and Machine Learning, Noor strongly believes that the world can be changed for the better using interdisciplinary technological innovations. She considers data the future of healthcare development around the world.
4. Timnit Gebru: Technical co-lead of the Ethical AI team at Google
Timnit finished her PhD in computer science at Stanford University. Her research area focuses on algorithmic bias and data mining. She is also the co-founder of the Black in AI group.
5. Rediet Abebe: Junior Fellow at Harvard Society of Fellows
Rediet is the first black woman to earn a computer science PhD from Cornell University. She is the co-founder of Black in AI and Mechanism Design for Social Good (MD4SG). Her research focuses on using algorithms to improve access to opportunity for historically marginalized communities.
The dangers of AI
As with any new technology, AI poses significant challenges, especially if in the wrong hands. Here are some of the dangers that come along with AI:
- AI-powered weapons
Just like we can train AI to help us do tasks that would normally take a long time to complete, we can train it to do heinous things. In fact, the nuclear arms race the world is in right now might easily be converted to a global autonomous weapons race. If the technology is used by the wrong hands, it will be difficult to dismantle or combat it after it is deployed.
- Social manipulation
AI algorithms have proven to be effective in targeted marketing campaigns. Thanks to the Internet, most big data companies have collected sufficient information about who we are, what we like, and can effectively summarize what we think. But this poses a serious danger when used in social and political contexts. For instance, in 2016, data from 50 million Facebook users was used to try and sway the outcome of the US presidential election.
- Invasion of privacy
Nowadays, it is possible to track and analyze an individual’s every move online as well as when they are going about their daily business. For example, China plans on using facial recognition system to power its Social Credit System which will give every one of its citizens a personal score based on how they behave. So things like jaywalking, smoking in restricted areas and time spent on video games all factor into the score a person will get. This is a huge invasion of personal privacy.
AI is trained on existing data, which means it is also learning all the biases we have in our data; so algorithms have shown to be racist, sexist and discriminatory. For instance, Google’s face recognition system broke headlines when it tagged two African Americans as gorillas. Google apologized for the mistake and claimed they fixed the algorithm while in reality they just removed gorillas from the system. AI assisted recruitment systems have thrown out CVs of women and minorities simply because of their nature. For instance, the algorithm that Amazon deployed between 2014 and 2017 to screen job applicants reportedly penalized words such as “women” or the names of women colleges.
While AI is great at learning from data and adapting to change, one thing it is really bad at is understanding context. In short, AI takes things too literally. So if you tell it to get you to the airport as fast as possible, it will get you there as fast as possible—by ANY means necessary. Following the traffic rules and watching out for pedestrians is not going to pop in the AI unless it had been trained to always follow traffic rules on the road. So through misunderstandings like this, a lot of damage could occur.
Is it monitored?
The public sector has actively engaged in the development of policies and laws that promote and regulate AI. While the research is important and many countries believe in its value, they also deem it necessary to manage the associated risks. Laws and regulations concerned with AI revolve around three topics of interest:
- Governance of autonomous intelligence systems
- Responsibility and accountability for the systems
- Privacy and safety issues
The regulations put in place are especially concerned with closely monitoring the development of Artificial General Intelligence (AGI), which refers to the hypothetical intelligence of a machine that has the capacity to understand or learn any intellectual task that a human being can. While AGI is risky, the public sector is even more concerned with preventing the development of a superintelligent AI, one that is capable of creating other superintelligent AI. In its current state, no AI can create another of its kind in full functionality. A superintelligent would be capable of over shining the brightest human minds, hence, there is global concern as to how it could be controlled.
Since 2017, the world has been calling for the establishment of a global government board to regulate AI development. In December 2018, Canada and France announced their plans for a G7-backed International Panel on Artificial Intelligence, similar to the International Panel on Climate Change. In 2019, the panel was renamed the Global Partnership on AI.
Many stakeholders have started a discussion about the regulation of AI. China has a regulation for AI called “A Next Generation Artificial Intelligence Development Plan.” The regulation is concerned with the development of AI and ensures state control of Chinese companies. It restricts companies in their use and storage of valuable data, including limiting the storage of data on Chinese users within the country and mandatory use of People’s Republic of China National Standards on AI.
The European Union is guided by a European Strategy on AI. In April 2019, the European Commission published its Ethics Guidelines for Trustworthy Artificial Intelligence (AI), followed by its policy and investment recommendations for trustworthy AI in June 2019. On February 02, 2020, the European Commission published its White Paper on AI—A European approach to excellence and trust. One of the building blocks of the White Paper, “ecosystem of trust,” outlines the EU’s approach for a regulatory framework for AI.
In the U.S., discussions on the regulation of AI have touched on topics such as timelines for regulating AI, the nature of the federal regulatory framework to govern and promote AI, how to update regulations in the face of rapidly changing technology, and the roles of state governments and courts. On January 7, 2019, following an Executive Order on Maintaining American Leadership in Artificial Intelligence, the White House’s Office of Science and Technology Policy released a draft Guidance for Regulation of Artificial Intelligence Applications, which includes 10 principles for U.S. agencies when deciding whether and how to regulate AI. In response, the National Institute of Standards and Technology has released a position paper, the National Security Commission on Artificial Intelligence has published an interim report, and the Defense Innovation Board has issued recommendations on the ethical use of AI.
Artificial Intelligence is not an abstract concept as many might think it is. It is a technology with the potential to transform the world for the better with applications ranging from agriculture and healthcare to luxurious ways of living in smart homes and owning self-driving cars. With its wide range of application and potential to make life on Earth easier, research in the field needs significant attention. As research and development in the field is essential, educating the general public about the use and development of AI is equally important, because people who may not understand what AI is will likely resist its applications. At the same time, the development of AI needs to be closely monitored as it is extremely easy to use it for malevolent agendas.
Share on your socials!