What can AI do for our society?

You wake up bright and happy, feeling a fresh wave of motivation preparing you to take on the day. While driving to visit your mum, your phone instructs you to make an unusual turn into a side street – as you did so, you couldn’t help but notice a long line of traffic grinding to a standstill on the main road.

An hour before finishing work, your phone tells you to leave 15 mins earlier than usual if you want to make tonight’s dinner because of train delays. It then tells you to enter London Bridge station via exit 2 and board the second carriage from the front. On your way out, it didn’t go unnoticed that your door aligned perfectly with the platform exit, allowing you to get out before the sea of people poured out of the tube and drowned the platform.

As you turned the corner of Piccadilly Circus, you couldn’t help but feel anxious about your choice of restaurant, put immediately to rest when you recalled seeing on your phone there’s an 89% likelihood of your partner liking this particular restaurant. Fortunately, dinner goes perfectly – perhaps a little too perfectly – and now you need a ride home.

No Big Deal.

You grab your earphone out of your pocket and summon a car through your smart assistant – your watch telling you it’s only 5 mins away. During your ride, you enjoy a carefree time with your partner, afforded by a level of privacy only a driverless car can offer.

This is only the beginning of what AI can do for us.

Waking up happy and fresh doesn’t have to be mere coincidence; you can use apps which prompt you to stay hydrated and provide suggestions about the nutrient value of the food you eat. Combine this with smart alarms which learns to track your body’s circadian rhythm and REM cycles – waking you at just the right moment so you leave your bed feeling alert. Each of the suggestions your phone made in the situation above required the system to make observations about the choices you make throughout your daily life. These observations allow a machine learning model to learn and make predictions about future choices that should be made to reach some kind of goal. In the case of travelling to the restaurant, the model’s goal is to minimise the time it takes to reach the destination. To that end, it will provide us with a set of choices we should make during our journey.

If you think AI is going to bring an end to independent thinking you wouldn’t be the first one to think so. However, I believe this perspective is quite myopic. Consider the fact that in today’s globalised society more and more choose to travel and work abroad. For a local Londoner, knowing which carriage to get on in the Tube is a matter of habit, but for the entrepreneur from Kuala Lumpur who’s on her way to a meeting in Moorgate, this is indispensable knowledge she didn’t have a chance to acquire, but now have access to. In this context, AI is helping make our world a smaller place – making the unfamiliar, familiar and the inaccessible, accessible.

The collection of personal data to facilitate these services still remain a large and controversial topic which I will elaborate upon in a future post. In the meantime, let’s explore what difference AI can make to three fundamental areas of society.

Healthcare

In 2019 the NHS announced plans to set up a national AI lab to help them deliver more personalised and efficient healthcare. For instance, through medical imaging, AI will be able to identify types of cancers and suggest treatment plans that will be responsive to a particular patient. People who are prone to missing an appointment can be given extra attention. This level of personalised and proactive care will be nothing short of revolutionary.

With intelligence comes the ability to democratise expertise. Systems can be operated by volunteers in remote areas where care is needed the most. Imagine volunteers in Bangladesh with tablets identifying patients with deadly disease in the Rohingya refugee camp from spreading further. Using this system, doctors can remotely diagnose patients from their office without having to lose time travelling around the world.

Currently, the sharing of data across different healthcare organisations remain a challenge. Before AI can begin solving our problems, governments will need to create policies which take into account the privacy of patients and facilitate data sharing through public cloud infrastructure whose operators do not stand to profit from selling patient data. The machine learning model training process will need to be representative of society and take bias into account as many research systems today use Caucasians in their training data, which has been shown to perform poorly on people of other ethnic groups.

Education

As we design better AI systems, we continue to learn more about the nature of the learning process itself. This can help teachers develop curriculum which take full advantage of the student’s learning abilities. Moreover, data about students can shed light about what works and what doesn’t, so educators can focus on areas of improvement down to individual student. This approach is so scaleable, it can be used in online learning platforms so that independent students can also benefit from virtual supervision.

Analytics can be used to predict which students are likely to fail or drop out from a course, so that help can be given to those students early. This can massively reduce mental health difficulties many students currently suffer from. Furthermore, models will be able to predict if students are likely to be frustrated or confused given some piece of content which can help any teacher understand ‘at-a-glance’ what the current status of a given student is – similar to vital stats medical professionals use to determine the condition of a patient.

Learning companions, similar to smart assistants, can be developed which guide and test students through any educational material. It can also synthesise the content, giving students the ability to digest much more than they normally could – all the while acting as a supervisor which can ask the student to take timely breaks and motivate them to success.

However, we must remember that technology can be no substitute for individual care. We can imagine as AI systems become more intelligent, teachers can grow to rely on it increasingly. This is not necessarily bad as it frees them up to research more interesting content and present it in a more creative fashion. But when it comes to teaching children, policy needs to be established which makes it clear to teachers that human care and attention must be given as this is rather important in the child development process. Also, data about students should be collected and stored in a responsible manner to prevent prejudice and discrimination. Potentially insidious is the use of learning companions as propaganda devices which act to skew an individual’s worldview with the intention of influencing their choices.

Public Sector

Governments collect data and this makes them a fertile ground for machine learning to be applied within many of its sectors. In the UK for example, fraudulent benefit claims are rife. To clamp down on culprits, models can draw data from various systems to build a profile on citizens which can then be used to red flag potential offenders. This promises to save the tax payers a lot of money as investigations will be targeted and thus, incur less labour costs. Infrastructure failures can be predicted and fixed before they happen, minimising massive disruptions and macro patterns in traffic could be used to inform improvements to existing roads.

Crime can also be tackled in a way that has never been done before. The Metropolitan Police has invested a lot in body cameras which are able to work together and stitch scenes of crime – offering a new vantage never seen before. On top of this, the use facial recognition has recently come into effect in London and can be used to identify dangerous offenders and escapees on the streets which can be used to bring a new level of safety and vigilance on our streets.

However, if not used in a responsible and ethical manner these same tools can be used to stifle progress and change in society. China, for instance, has come under criticism by privacy and civil liberty advocates for using intrusive technologies such as applying facial recognition to toilet tissue dispensers in public toilets and the social credit system which restricts the ability of citizens to get a mortgage or buy a flight ticket based on their behaviour. Particularly controversial is the use of AI surveillance tools on Uighur people to carry out predictive policing á la Minority Report’s Precrime division to catch terrorists. However, in situations like these, we must be aware that machine learning models cannot be a panacea. They cannot help us identify terrorists unless we show it examples. And if holding a public prayer is defined as terrorism, then this technology will act as the great oppressor of people, turning society into a panopticon.

AI is a necessary tool to tackle the grand challenges of the 21st century. It will open many opportunities for fundamental change within all parts of our society and it is up to the collective to voice their opinion and influence its implementation in a desired way. The biggest concern we should have about AI are not sentient machines rebelling against humanity, but policies which give rise to autonomous weapons with the capability to eliminate individuals with no humans in its decision loop.

Related posts:

Those who trust in the Lord need not fear death. Our maker has conquered the keys of the grave. We have a future beyond this world. The time will come, not too far from now when the dead in Christ…

when my instructor announced this project, i was ready to bounce outta this class. i thought i was here to learn design, not how to draw black squares. i mean i thought we learned this in preschool…