20 Aug Do the Benefits of Artificial Intelligence Outweigh the Risks?

AI is not just about robots and the stereotyped perceptions etched in our minds

– Silka Gonzalez, President, Enterprise Risk Management

Coral Gables, Fla. – Aug. 20, 2018

This story originally appeared here

We live in a world where robots have begun getting citizenships. Yes, you read that right! Sophia, a humanoid robot was granted citizenship of Saudi Arabia recently. Sophia uses artificial intelligence and imitates human gestures and facial expressions, and takes part in various summits around the world where she communicates and gives intelligent responses to questions; much like humans do.

But Artificial Intelligence (AI) is not just about robots and the stereotyped perceptions of AI etched in our minds thanks to mass media. Plainly put, AI involves programming machines in a way that they are able to think on their own like we humans do. AI is all about impersonating human intelligence. This might sound like rocket science but in reality we’re already using AI in our daily lives a lot of times. Think Siri on the iPhone or how Google knows which emails need to be sent to the spam folder and which shouldn’t. AI is deeply integrated into our daily lives.

AI In Action

• Self-driving cars – These driverless vehicles are the latest revolution in the history of car making. These cars have computers in them that are programmed to make the best possible decisions in real-time driving situations. The computers are fed with massive amounts of data extracted from real-life situations.
• Google Maps – Map applications, like Google Maps, have the ability to calculate the fastest and shortest routes to our destination by analyzing the speed of traffic, incidents, real-time updates from drivers, and lots of other data crunching. Google has even come up with a feature to find you a parking space¹nearest to you. Well, that’s a relief!
• Virtual Assistants – They’re like real-life assistants and they’re free of cost! Apple’s Siri, Microsoft’s Cortana, Nina and a host of other such virtual assistants have popped up on the smartphone scene. But they’re not limited to smartphones alone. You might have come across similar virtual assistants on online websites as well where the site offers the customer a “live chat service” to help answer your questions. They’re called “Chatbots” and they’re gaining popularity on several websites.
• Video Games – AI isn’t new in gaming. But the complexity levels have risen to such an extent that gaming characters today can learn your behavior, understand your preferred style and act/react to these inputs to give you a more individualized gaming experience.
• Autopilot in Flights – This is one of the earliest uses of AI and today it’s actually used for a majority of all flight journeys. In reality, the human pilots today, usually, operate the controls manually only during takeoff and landing. All the other time, it’s autopilot that’s flying you around the world.
• Smart Appliances – Consumer electronics implemented on AI technology are being used widely today. Televisions, refrigerators, thermostats, bulbs, security cameras, health monitors and lots of other gadgets, under the collective banner of “Internet of Things (IoT)”, are popular mainstream products today.
• Spam Filters in Email – Spam filters often use an artificial neural network to detect and block spam emails. Gmail, reportedly, has a track record of filtering out 99.9% of all spam emails using AI technology².
• Industry and Cybersecurity – While robotic arms and even stereotypical robots have been in use in industries like automobile manufacturing, artificial intelligence is today find its way into almost every industry you can imagine. Even the cybersecurity industry has seen and continues to see improvements and advances in the use of artificial intelligence to thwart hackers and cyberattacks.

It’s beginning to feel like every area in the business world is soaking in AI. And researchers are constantly on the lookout to identify other potential areas where AI could be adapted.

Benefits of AI

• Efficiency and Cost Effectiveness: AI often employs highly sophisticated underlying technology and computing machinery that have the ability to quickly analyze vast amounts of data. Therefore, jobs that are monotonous in nature can be done with greater ease, speed, and accuracy. All of this is also made available at reasonable costs and, sometimes, no cost at all.
• Life is Easier: People rely on various AI based applications today to make their lives easier. A thermostat that knows when you get back home and adjusts the temperature accordingly, lights getting dimmer and brighter depending on your location in the house, digital assistants like Alexa and Siri that help you in various ways like setting alarms, reading books, finding restaurants and much more.
• Medical Purposes: AI has found very valuable applications in the medical domain. It has made diagnostics more accurate than ever. Doctors have started using robots and robotic arms to perform specific types of surgeries for more precision work. In fact, remote surgeries are now possible where the doctor might be located in New York but performs surgery on a patient in Paris with the help of robotic arms and Internet connectivity. Surgery simulators are used today to train medical practitioners. Therapeutic robots are available to improve the quality of a patient’s life and wearable health trackers like Fitbit monitor your health on the go. These are only a few example; the list is truly long.
• Overcoming Human Limitations: AI can reach places and help human do things that we’re otherwise limited at doing. For example, exploring space to greater extents where it could be dangerous for humans to reach or reaching the deep ocean where humans might not survive.
• Intelligent Predictions: Technological advancements have created a sea of data. This is one of the main reasons that AI is being used in various institutions, especially the medical and financial domains, to organize and manage data. AI can then analyze data to find patterns, extrapolate based on these patterns, and arrive at predictions that can enable better and faster decision making.

Disadvantages of Artificial Intelligence

There’s no doubt that AI holds great promise. However, getting tasks done by machines with little or no human intervention comes with its own set of challenges and disadvantages.

• Health Issues: Connections with people are turning digital. Worldwide, there 2.13 billion monthly active Facebook users and 5 new profiles are created every second³. Facebook, Instagram and other social networking sites use AI technology to make these platforms more addictive for you. People are getting obsessed with this excessive usage of social media and it negatively influences their health and well-being. Anxiety disorders, depression, and sleep deprivation are just some of these negative effects.
• Environmental Toll: The tremendous amount of data being created and moved around is actually one of the reasons for global warming, which leads to various kinds of disturbances in weather and weather patterns⁴. We’d probably do good to slow down and analyze the data footprint we’re creating and how it impacts our natural resources.
• Unemployment: AI could lead to unemployment if machines end up taking over jobs that humans do. Since humans cannot work 24/7 and machines can, industries are drifting towards the use of AI technology to get seamless work day in and day out. What’s more – it can also help reduce costs.
• Artificial Idiots: From a cybersecurity standpoint, AI throws up some serious questions. You’re essentially leaving systems and entire processes, at times, on autopilot. Systems are fed with huge amounts of data so that they can learn trends and formats from that data and act accordingly. But what if a malicious hacker or insider fed in malicious and directed data to make the AI infrastructure think and behave in a particular way? The cybersecurity world has often talked about how humans are the weakest link. In situations like these, computers too become the weakest link.
• AI in the Wrong Hands: If machines fall into the wrong hands they could be used to commit new-age terrorism, automated cyber-attacks and other vicious things limited only by the extent of a malicious individual’s imagination. Researchers have found ways to manipulate a speaker’s speech into new speech that’s also lip synced. So you could actually make someone say things they never actually said and that too in their own voice tone⁶. Fake news would hit record levels.

Threats

With the wide adoption of AI in various sectors, cyber criminals are already shifting their focus on AI. Symantec recently predicted that in 2018 cyber criminals will use AI to launch more sophisticated attacks⁵. And AI by itself, comes with some serious cybersecurity questions that need asking. Here is a glimpse of some of the threats that AI poses.

• Too Much Data: AI has the capability of working on enormous amounts of data. The data collected is not just simple data but rather data that includes every minute detail about an event or transaction. This is also known as “Big Data”. To imagine a situation where a hacker gets hold of Big Data should be a cause of serious concern. Also since we’re talking about automating the decision-making process, there is no transparency as to why the machine reached at a certain conclusion and since there is so much data involved, it becomes technically infeasible for humans to understand what’s going on. In fact, chances are high that humans will end up missing the discrepancies in the results shown by machines. Think of this in a critical infrastructure related industry setup like water treatment plants, nuclear reactors, rail transport, and so on, and you’ll see the magnitude of the threat at hand.
• Data Privacy: A large number of AI-enabled devices today gather a significant amount of personal information on individuals. Take the example of IoT gadgets – health monitors, fitness gadgets, smart televisions, and a whole host of other IoT gadgets gather a lot of information on individuals and even the environment around them. This data can be very revealing in the wrong hands. A lot of the new-fangled devices do not address key cybersecurity questions and concerns such as – where is the data being stored, what protection measures are being used to protect the data, who controls this data, is the data shared, was consent taken, and other such difficult but critical questions. Given that these devices are today part of an individual’s daily life, it affects the organization where the individual works as well and changes the whole dynamic of the way cybersecurity needs to be approached in organizations.
• AI – The New Hacker: Hackers too have started using AI as a powerful new tool. Recently, two data scientists conducted an experiment to identify who between humans and AI who was better in making Twitter users click on a malicious link. The scientists taught AI to study the social network user’s behavior and then design and implement its own phishing posts⁷. Turns out that the AI hacker’s performance was much better than the human hackers. There are ominous signs in here that evidence exactly how innovation is a double-edged sword.
• Out of Control: Machines/Robots based on AI are created with the ability to make intellectual decisions like humans. In fact, AI is taught to “learn” on its own as well to be able to take informed decisions even if it never had the data related to similar situations. If an intelligent capability has been designed in a self-learning manner, it’s probably not far-fetched to say that there is the possibility of the intelligent machine developing a “mind of its own”. If not by itself, on could argue that with the “assistance” of a malicious hacker, such a situation could be brought about. Such an AI capability remotely controlled by a malicious hacker could truly wreak havoc.

Final Words

AI is a field full of promise and potential. But like almost all new innovative domains, it is in a nascent stage and everything that’s coming out of its doors is unregulated. The threats associated with AI are very real as well and it would be ill-advised to wait for cybersecurity regulations to come in and save the day. In any case, regulatory compliance should be a by-product of good cybersecurity measures and infrastructure and not the other way round. AI is truly exciting, but humans should probably pause and think about where they intend to draw the line.

– Silka Gonzalez, CPA CISSP CISA CISM CITP QSA, is President at Enterprise Risk Management

Enterprise Risk Management (ERM) is Cybersecurity company that specializes in Cybersecurity services and training. ERM’s services include Cybersecurity strategy, assessments, remediation, implementation, digital forensics, products and cybersecurity culture and awareness training. ERM has served over 300 clients in over 20 industries. Clients range from private and publicly held multinational companies to small businesses. Sample industries include banking, financial services, education, government, healthcare, retail, and technology.

© 2018 Enterprise Risk Management, Inc. All rights reserved.

EM Risk Archives