Artificial Intelligence and Machine Learning techniques have actually been around for some time. It is only more recently that we are hearing more about them in our daily lives.
Artificial Intelligence, or AI, conjures up images of the Terminator, a sentient cyborg and malicious AI creation from the future, sent back to the past to take over humankind, or Robocop, an ethically questionable hybrid human-cyborg created to enforce the law in a lawlessness city.
Science fiction has always been the gazing lens for theatrical dystopian futures when it comes to artificial intelligence, and although great minds such as Stephen Hawking caution against the potential future of robotics surpassing human intelligence, and 'taking over', there is also a lot of good that can come from this area of computer science.
For example, facial recognition built in to smartphones, virtual assistance such as Siri and Alexa, and the health sector, where I have spent all of my career to date, is seeing some fantastic, innovative, initiatives such as improved speed and accuracy of medial diagnosis.
However, although the science fiction future is likely exaggerated, we do need to be cautious when developing artificial intelligence and machine learning techniques because like all other technological evolutions, if they fall in to the wrong hands, a lot of damage can be done.
In this article, we look at what Artificial Intelligence and Machine Learning is, including their closely related technologies such as deep learning and neural networks.
We then look at the benefits these techniques bring to improving our online safety, and the risk if they fall in to the wrong hands.
AI is the culmination of underpinning techniques such as machine learning and deep learning, where systems, robots or other types of machine, imitate human behaviour.
Our behavioural traits include, for example, the ability to plan, learn, problem-solve, acquire knowledge and draw conclusions based on reasoning applied to the information we have available. In other words, the ability to 'think like a human'.
Our physical traits include, for example, the ability to walk, run, jump and twist. In other words, the ability to 'move like a human'.
Combine the two and you have Artificial Intelligence. The underpinning techniques that conclude in an AI innovation are Machine Learning and Deep Learning.
AI is experienced in everyday life these days. Think about Siri or Alexa, and their ability to recognize speech and interpret the natural language human used to communicate, and to provide you with a response or answer.
What about sat nav's calculating the most efficient route to your destination based on big data such as traffic volume, traffic flows and road works?
This is collectively known as Assisted Intelligence. It is the basic form of AI that supports human planning and enhances the activities humans are undertaking during their AI interaction, e.g. "Siri, play me some blues music", or "Alexa, who plated the lead role in Blade Runner?"
The next step up is referred to as Augmented Intelligence. This is where AI enables humans to do things they otherwise couldn't do.
For example, reducing medical diagnosis errors, increasing diagnosis rates (and therefore earlier treatment plans, which is good for diseases such as cancer) and improving the precision of hip replacement cuts through robotics programmed to make cuts based on an individual's X-Rays and MRI scans.
This technology is emerging and will be part of our everyday lives before we know it. Then there is the next-generation AI, referred to as Autonomous Intelligence. This is where machines and systems imitate human intelligence and creativity, so they can 'take over' some tasks we undertake today.
For example, self-driving vehicles with the intelligence to avoid accidents, and the Mars rover than can negotiate the rough terrain of the planet's surface.
Machine Learning is the engine that fuels Artificial Intelligence. It is a set of mathematical algorithms programmed using languages such as C++ or Python, to consume and rapidly process huge quantities of different data sets and analyse that data to identify, or learn, trends, routines, patterns or anomalies humans would otherwise miss or be incapable of identifying.
The learning element is where the algorithms change the AI behaviour based on the latest data sets it is consuming. This is how AI can adapt to its changing environment and circumstances, "like a human would, only better".
The algorithms that fuel Artificial Intelligence and Machine Learning can be grouped into two different types of knowledge learning, based on their intended functions.
This is Deep Learning and are the cornerstone of Machine Learning which make the self-driving cars and medial diagnose techniques possible. :-
These techniques are set to take technology to the next level. Even Elon Musk is looking towards AI implants, enhancing humanity and fix problematic neurological conditions. The benefits are almost endless.
Before we move on to the next section which looks at the wider benefits of AI, take a moment to watch the short video below, which covers what Artificial Intelligence is in more detail.
The ability of Artificial Intelligence and Machine Learning to detect patterns of activity within huge and complex data sets is being incorporated in to the next generation cyber security tools.
For example, some modern malware activities are covert and infiltrate at a low level that would not otherwise be detected by traditional anti-virus or anti-malware solutions.
AI can detect and alert to this suspicious activity, informing you the home computer user and through the ability of cloud computing, share this information so that others using the same security software are also aware and their local anti-virus or anti-malware application can "keep an eye out" for the same suspicious activity.
This is fantastic for combating zero-day exploits, i.e. new malware exploiting exposed software vulnerabilities.
In addition, the algorithms can rapidly analyse the events from millions of devices around the world and alert to risky behaviour that could lead to a malware event such as a phishing attack or ransomware downloads, effectively implementing proactive preventative techniques.
As the body of knowledge grows, the patterns of suspicious online activity can be profiled, so any deviation from expected behaviours can be rapidly detected and acted upon.
In the modern world of cat and mouse between cyber security firms and the cyber criminals, it is imperative all new technologies are exploited for maximum benefit. The majority of us don't want our online experiences disrupted by malware, denial of service attacks or virus infections. This is where AI can help.
Artificial Intelligence and Machine Learning techniques are also being incorporated into home security systems such as CCTV and intruder alarms.
In addition, the home can be increasingly customized to suit your living style and habits, such as lighting and temperature control, and sensors to manage your goods and health. These "Smart Homes" are set to increase in popularity and will become the norm before we know it.
Google has been using AI for years, including facial recognition and voice detection techniques. Facial recognition has been used by police departments and federal bureaus, although there are some questions around the regulation of use.
The benefits of AI will ultimately help keep all of us a little more safe and secure online. There are too many instances of ransomware or fraudulent attacks for my liking.
However, what is stopping criminals from using this technology to their advantage? This is the focus of the next section.
As we touched on in the previous section, there are many questions to debate on the ethics of using Artificial Intelligence and Machine Learning techniques, such as those used across social media to influence widespread thinking, government agencies and marketing.
Deepfakes, is a technique whereby authentic images or video footage is manipulated to appear as if it is someone else. This can appear so realistic, there is an ethical question around what we can interpret as real or fake. Is it any wonder we don't know what to believe online?
Take a look at the two deepfake videos below. The first one of the Queen was aired in the UK on Christmas Day 2020. It is amusing, and you can see the manipulation is not quite real (intentional or not?).
The second video, again amusing, is more of a warning on what is going to be possible in the future.
Also, can we be certain the data used to fuel AI will remain safe, and will the conclusions and courses of action from that data be legal and ethical?
As mentioned earlier in the article, the work of the algorithms and their conclusions reflect the quality of the data they receive. What is to stop hackers using the same AI techniques to manipulate that data?
What is stopping hackers using the same Machine Learning techniques to avoid detection and perform their malicious intents? What if AI can be used to develop malware that mutates according to the environment it is operating in, "COVID-like"?
Cyber security firms will need to develop new ways of combating these new techniques when used for malicious purposes.
What's to stop them being used in your future Smart Home to disable intruder detection systems, or switching off your smoke alarms and human detection sensors, before burgling your home and burning it down?
I think it is fair to say that Artificial Intelligence and Machine Learning are going to bring fantastic benefits in to our daily lives. I like the idea of a secure Smart Home, and driverless cars will be a lifesaver if done correctly.
However, it is also fair to say that AI and ML are not a panacea for all of our cyber security and malware issues. The algorithms can be manipulated and therefore used maliciously. This needs to be avoided at all costs before AI and ML become mainstream.
Finally, take a look at the AI for Business article by ZDNet, which is littered with example of AI and ML uses and covers a multitude of AI aspects.