Interview Questions for Ben Luria, Founder and CEO of Hirundo
In the rapidly evolving world of artificial intelligence (AI), ensuring safety, reliability, and adaptability has become a paramount concern. Enter Hirundo, an Israel-based company that has developed a groundbreaking solution: the Machine Unlearning platform.
The core challenge in fixing problems in trained AI models is that it's costly and time-consuming to selectively remove specific information or behaviours. However, Hirundo's engine uses a unique "neurosurgical" method to pinpoint and alter specific values in the model's weights and vectors, making the process more efficient.
This approach, which focuses on redaction rather than addition, is a departure from traditional methods. It allows Hirundo's platform to quickly and effectively erase specific information from large language models (LLMs), reducing hallucinations, biases, and security flaws.
The platform has proven to be highly effective, decreasing hallucinations and biased responses by more than 55 percent, and reducing jailbreak vulnerabilities by up to 85 percent. This makes it particularly valuable to teams working on mission-critical, high-risk, or regulated AI systems.
Business leaders should think of machine unlearning as a tool that transforms AI development into a more controlled adjustment, making models safer, more reliable, and easier to align with evolving requirements.
Hirundo's Machine Unlearning platform is already in use by enterprise LLM and data science teams, Responsible AI and AI Safety teams, and frontier AI labs. Notable companies such as NICE and Check Point in Israel are using Hirundo's technology to make their AI systems safer and more reliable.
Moreover, the platform can remove personally identifiable information, confidential knowledge, and toxic behaviours without retraining. This means that businesses can adapt to changing regulatory and security needs quickly, without the need for extensive model retraining.
Over time, machine unlearning creates AI systems that remain adaptable and trustworthy as business, regulatory, and security needs evolve. By keeping pace with society's expectations, machine unlearning lays the foundation for AI that can scale responsibly without sacrificing safety or control.
Ben Luria, the CEO of Hirundo, is at the helm of this revolutionary technology. His vision for a safer, more reliable AI future is becoming a reality, one redaction at a time.
In conclusion, Hirundo's Machine Unlearning platform is a game-changer in the AI industry. By focusing on redaction rather than addition, it offers a more efficient, effective, and controlled approach to fixing problems in trained AI models. This not only improves overall model stability but also makes AI systems safer, more reliable, and easier to align with evolving requirements.
