EXECUTIVE INSIGHT
However , humans are still essential for training these models . Poor data leads to poor outputs and results . Therefore , companies must understand their AI objectives and employ experts capable of accurately defining AI models .
In the past , I ’ ve witnessed instances where people developed a malware detection model without understanding how the malware operates . Consequently , the model identified incorrect artefacts from the malware , resulting in numerous false positives .
Q . LOOKING AHEAD , WHAT TRENDS DO YOU ANTICIPATE WILL HAVE THE MOST SIGNIFICANT IMPACT ON THE FIELD OF THREAT ANALYTICS IN THE COMING YEARS ?
» The challenge in threat analytics lies in striking the right balance between gathering necessary data and addressing data privacy concerns .
We are increasingly engaging in conversations with customers about data privacy and protection . It ’ s a conversation we very much welcome , and one that the entire security industry should be having .
I ’ m also excited about the emergence of AI-specific CPU architectures . They will enable significant advancements in processing power , enhancing capabilities in data analytics and data virtualisation on the defensive side of security .