AI CAPABILITIES FOR SOCIAL GOOD
The world we are living in today feels like a wonderland. Have you ever wondered why you get ads based on what you say or touch? Sometimes it might be what you watch on tv and then the next time you log into your social sites, you start seeing different ads based on your view. Well, that happens to me all the time. Sometimes I may watch a particular tv show or search for a product and then all of a sudden I’m seeing an ad on my social media for that same product. Do not wonder anymore because that is the capabilities of Artificial intelligence and I will be telling you how it works in this article.
First, what is Artificial Intelligence?
According to Wiki and leading AI textbooks, Artificial Intelligence(AI) is an intelligence created by a machine. It is any system that perceives its environment and takes actions that maximize its action of achieving its goal. Artificial intelligence is a wide-ranging branch of computer science concerned with building smart machines capable of performing tasks that typically require human intelligence.
Types of Artificial Intelligence and its example
- Reactive Machines
This follows the most basic of AI principles as its name implies. It is capable of only using its intelligence to perceive and react to the world in front of it. A reactive machine cannot store memory and as a result, cannot rely on past experience to inform decision-making in real-time. An example of where reactive machine AI is being used is a game-playing reactive machine Google’s AlphaGO. AlphaGo game is incapable of evaluating future moves but relies on its own neural network to evaluate developments of the present game, giving it an edge over Deep blue in a more complex game.
2. Limited Memory
This has the ability to store previous data and predictions when gathering information and weighing potential decisions, which essentially looks into the past for clues on what may come next. Limited memory artificial memory is more complex and presents greater possibilities than reactive machines. When utilizing limited memory AI in machine learning, six steps must be followed: Training data must be created, the machine learning model must be created, the model must be able to make predictions, and the model must be able to receive human or environmental feedback, that feedback must be stored as data, and these steps must be reiterated as a cycle. Examples of where Limited Memory AI is being used are mostly our social media, online stores, etc
3. Theory of Mind
Theory of mind is theoretical. We have not yet achieved the technological and scientific capabilities necessary to reach this next level of artificial intelligence. Its concept is based on the psychological premise of understanding that other living things have thoughts and emotions that affect the behavior of one’s self.
4. Self Awareness
Once the Theory of mind can be established in artificial intelligence, maybe in the future, then the final step will be for AI to become self-aware. It basically relies on human researchers understanding the premise of consciousness and then learning how to replicate that so it can be built into machines.
In regular Intervals since the 1950s, experts predicted that it will only take a few years until we reach Artificial General Intelligence — Systems that will show behavior indistinguishable from humans in all aspects and that AI will also have cognitive, emotional, and social intelligence. In 1956 Marvin Minsky and John McCarthy, a computer scientist at Stanford, hosted the approximately eight-week-long Dartmouth Summer Research Project on Artificial Intelligence(DSRPAI) at Dartmouth College in New Hampshire to prove the prediction. This workshop marks the beginning of AI. Today, Artificial Intelligence(AI) is one of the most rapidly expanding areas in the technology industry. The use of AI has extended to a wide range of industries, including healthcare, transportation, speech recognition, self-driving cars, social media, and security. It is in many ways an important source of change in people’s lives in the 21st century. Also, AI has shown success in games and simulations and is being increasingly applied to a wide range of practical problems. These commercial applications often have indirect positive social impacts by increasing the availability of information through better search and language-translation tools, providing improved communication services, enabling more efficient transportation, supporting more personalized healthcare, recommendations on social sites, etc. With this interest come a lot of questions regarding the impact of AI and how it can be used for social good.
What exactly is AI for social good?
AI for social good is a vague concept, given that neither AI nor social good has a widely accepted definition. As Bettina Berendt points out, in AI for common good thesis social good is a common good that is referred to as a goal, which is yet to be defined. He also stated that AI should be beneficial to both the human and natural environment. Meanwhile in the past, the AI researcher community (AI4SG) — “AI for social good” aimed at developing AI methods and tools to address problems at the societal level and improve the well-being of the society. Over the years, there have been several successful AI4SG projects such as guiding municipal water pipe replacement, protecting wildlife poaching, spreading HIV prevention information among homeless youth which has really made an impact in the society, and many more.
Impact of AI on social good
Artificial Intelligence in society is drastic and real. For example:
- Youtube’s AI-driven recommendation system would present sports videos for days if one happens to watch a live game on the platform
- Email writing becomes much faster with Machine Learning(ML) based auto-completion.
- Businesses have adopted natural language processing based chatbots as part of their customer services
- Games such as Poker use AI
- Android and IOS now use AI for recommendations and personal ads.
- Social media uses AI for personal ads etc.
How AI capabilities are used for societal benefit
AI capabilities like computer vision, natural language processing, and the capability of structured deep learning had the most widespread and greatest potential for a social good application.
- Within Computer vision, the specific capabilities of image classification and object detection stand out for their potential applications for social good. Image classification is an AI capability that classifies an Image or video clip into one of a set of predetermined classes. For example, Image classification can tell you whether the reviews contain a cat or a dog. It can be used to solve problems relating to crime, and poverty levels of different neighborhoods with daytime imagery. Also, Object detection is a capability of AI that can find all the instances of all trained classes of objects and report their locations within an image. For example, it can be used to solve problems such as the detection of fires in satellite imagery and the detection of bacteria in plants.
- Natural Language processing can also be applied for societal impact where language and communication barriers are a roadblock. This includes sentiment analysis, language translation, and language understanding. It is most useful in unstructured text form, documents, incident reports, articles, health records, and social media posts such as tweets. For instance, language translation provides value where language and communication barriers are major roadblocks. Also, it can be used to detect distorted information and combat fraud.
- Structured deep learning(SDL) analyzes traditional tabular data sets, which are often accessible for societal impact uses. It can contribute to solving problems ranging from identifying fraud based on tax return data to finding patterns of insights in electronic health records that would be very hard for humans to discover. For now, the use of structured deep learning has started to emerge in commercial sectors. For example, Instacart uses SDL to enable its shoppers to efficiently navigate stores, while reducing delivery time. Also, Pinterest developed an SDL system that helps to increase its recommendation related to pin engagement by five percent.
- Sound detection and recognition is also a common capability of AI that can be used for social good. Today, it is used for diagnosis in health, maintenance of transportation, and providing learning to students in education.
- Reinforcement and content generation is another capability of AI for social good. It involves learning by trial and error. It can be used for example to train models that can recommend precision medicine-based treatments for individual patients.
The Image below analyzes how AI can be used for social good and its use cases
Also, researchers and Social Sector experts mention four categories of limitations to AI use. See the image below:
The development of AI for social good will probably offer additional chances to enhance the community and unlock the opportunity for positive impact but there are risks that need to be managed. That said, there are several articles that focus on AI for social good if you would like to read more or want more information. I will be listing a few as references.
Feel free to explore!
CCC -Computing Community Consortium https://cra.org/ccc/wp-content/uploads/sites/2/2016/04/AI-for-Social-Good-Workshop-Report.pdf
Zheyaun Ryan Shi, Claire Wang, Fei Fang — Artificial Intelligence for Social Good: A Survey
Mckinsey Global Institute — Notes from the AI frontier applying AI for social good