In the biggest security event of the year, the RSA Conference (RSAC), many companies showcased their artificial intelligence (AI) technologies. It was evident that the use of AI-based products has become very popular and widespread, but not all experts use the terminology correctly.
One such AI-based product that grabbed a lot of attention is ChatGPT, which is an ML-based chatbot. Its technology is used to gather data points from various sources and combine them to create a usable model. The RSAC provided a platform for discussing and exploring various AI use cases and applications.
One of the most important areas in which AI is being used is the hiring process. It is not possible for the HR team to manually review all resumes and interview all candidates. Therefore, sorting the wheat from the chaff and creating a meaningful vetted shortlist is a task managed by an ML model, allowing the managers to review candidates’ shortlist easily.
However, the potential bias of the ML model is a significant concern that may affect the selection process. The model might have been trained on biased input data, which may create issues during the hiring process. Therefore, the use of this technology is considered an imperfect tool, but still much better than text searches conducted by humans.
Another essential issue is the possibility of bad actors infiltrating a company’s development environment. It is challenging to keep track of all the development tool chains in real-time, especially when third parties are involved. An ML-based reputation monitoring solution can help companies detect a breach of their development environment by third parties in real-time.
The issue of Deepfakes is another promising area, which was explored at the conference. Deepfakes refer to the use of AI to generate fake content, such as photos, videos, and audio clips. In some instances, Deepfakes can be used to create fake CEOs’ videos, and a money transfer request from them can cause irreparable damage to a company. Therefore, the use of ML can be an excellent solution for detecting deepfakes to avoid potential threats.
Privacy is a major concern in an AI-powered world. With the vast amount of data available today, many AI models require access to sensitive information, making it crucial to ensure privacy standards are in place. At RSAC, one of the startups presented a way to make data to and from ML models private by using interesting coding techniques. This solution is one of many attempts to address the massive set of challenges that are inherent in large language models forming the foundation of well-trained ML models.
The threat of insecure code in a rapidly changing threat landscape is also a significant concern, especially when integrating cloud properties that may be insecure. Detecting vulnerabilities in the early stages of development can save a lot of time, money, and effort in the long run. Therefore, many companies are now looking to ML algorithms to help identify vulnerabilities in software development before they can be deployed.
Despite the numerous benefits of AI, there are some serious concerns that have arisen from its use. With the power of early models, it is easy to imagine the fright and uncertainty over how powerful they can be and how they can be used. Therefore, there is a moral imperative that needs to be discussed and sorted out within the technology context.
In conclusion, AI-based products have become a crucial part of the technology industry and are being used in various applications. While there are still concerns over their bias, privacy, and security, the benefits of using them in various industries such as HR, development, and cybersecurity cannot be ignored. It is up to the practitioners in the industry to continue to evolve and improve the AI-based products in these areas, keeping in mind the moral imperative.