Tesla is holding the annual Tesla AI Day 2022 today, where the company showcases its innovations and plans from various business units to the public. In previous years, Tesla announced its Cybertruck electric light van, Optimus, a humanoid robot, and future-proof products such as advanced autonomous driving technologies. The innovations presented at Tesla AI Day events demonstrate how machine learning and artificial intelligence systems can be used in robotic or autonomous vehicles.

Elena Krupenina, Kaspersky Senior Data Scientist

For solutions based on artificial intelligence, cybersecurity measures are at least as important as the effectiveness and usability of these solutions. To ensure the safety of its users, an AI system must consider areas such as opaque algorithms and privacy, which can have unexplained consequences. So why?

Most machine learning (ML) models that support complex AI systems produce results or actions beyond human interpretation. While these results or actions may be unexpected, they may also lead to uncertainties. An example of this is when a robot catches one object instead of another due to confusion. Unexplained results can potentially pose risks to the system itself and to people. Therefore, developers should consider mechanisms and tools to evaluate and explain AI’s opaque decisions, then calibrate their parameters and metrics to avoid risk.

AI-powered devices can use a variety of sensors, such as cameras, microphones, radars, lidars, ultrasonic sensors, infrared cameras, to collect data. Huge amounts of data are needed to run AI at high quality, and minimizing the risk of exposing sensitive data is critical. To solve this problem, technologies must be developed that increase accountability and privacy, such as on-device processing and unified learning, which minimize the data required for the model to function correctly.

Source: (BYZHA) – Beyaz News Agency