Artificial intelligence (AI), one of many double-edged swords in technology, is expected to incite both positive and negative impacts on people and industries worldwide. It’s constantly evolving and as technological advancements emerge, these changes are leading many people to question whether or not the benefits of AI outweigh the risks.
Today’s hot topic of AI, also known as narrow AI, is designed to perform and focus on a specific task such as facial recognition or playing chess. Apple’s Siri is another form of narrow AI and features speech and image recognition. While the name downplays its capabilities, narrow AI is more technical than it sounds. It can decipher patterns from data and outperform humans at these specific tasks. On the other hand, narrow AI is not capable of outperforming humans on a cognitive level, meaning AI-powered devices cannot empathize or use logical reasoning to develop an experiential understanding of the human environment.
Further down the road we expect strong AI, the self-aware version of AI, to emerge. This level of technological advancement is not yet available, but this type of machine intelligence is expected to successfully perform, and possibly outperform, any intellectual task of a human. To many individuals, this notion of strong AI is extremely daunting. The silver lining for those with doubts, however, is that strong AI will not be available for decades to come.
AI has the potential to improve our lives, offering advances in healthcare, transportation and logistics, but it also imposes challenges on the cybersecurity industry. Today, malicious actors, including nation-states, are using AI to undermine cybersecurity in foreign nations and businesses. Max Heinemeyer, Director of Threat Hunting at Darktrace, a company using AI to identify and combat cyberattacks, said narrow AI will supercharge malware for years to come. Human attackers using malware are attempting to mimic normal network behaviors and move around networks while avoiding detection. Humans are currently required to manually invade a network and learn the normal behaviors, but once AI is programmed to autonomously perform this task, it can be localized to fit any environment. Under these circumstances, the malware would be even more difficult to detect.
In the future, we could see AI driving advanced phishing attacks. Attackers may use compromised data alongside AI to create contextualized messages that enter existing email conversations. This type of attack would make it nearly impossible to distinguish a genuine email from a malicious email.
To combat AI-enabled challenges, many security professionals are advocating to fight fire with fire. Dave Palmer, Director of Technology at Darkface, believes cyber AI is the most effective solution that can be employed to counter threats in real time. AI-based cyber solutions can leverage algorithms to distinguish between normal and abnormal threats and also autonomously respond to threatening behavior. On the downside, a ProjectWise report revealed that AI-enabled security solutions have provided significantly more security alerts and false positives compared to other solutions. In its early stage of adoption, AI-enabled solutions have also delivered inaccurate results. Twenty-five percent of respondents said they do not plan on implementing other AI-enabled security solutions as a result of their experience.
There is no such thing as a perfect security solution. Additional research is required for AI-enabled solutions to reach maturity, and existing solutions show that there is a long way to go. It is unknown what the future of AI holds, but it is guaranteed to directly impact security measures in every industry.