Artificial Intelligence is experiencing explosive growth in 2023. The stock prices of technology giants such as Nvidia have been affected by this surge, which shows AI's growing influence on the technological landscape. This rapid progress has not come without its challenges. Model drift, model collapse, and hallucinations are all issues that continue to threaten AI safety and reliability.

AI market growth and Nvidia’s ascent

AI has become an integral part of our everyday lives and has transformed many industries. According to predictions, the market will grow by 20 times and reach $2 trillion in 2030. Nvidia is a market leader in the semiconductor industry and has been leading this growth. In 2023 its stock price rose by 179%, outperforming the Nasdaq Composite Index’s 22% increase.

Nvidia has proven expertise in accelerated computation, especially with its high-end graphics processor units (GPUs), which is vital to many sectors, including automotive, healthcare and gaming. ChatGPT by OpenAI, for example, and Nvidia's expertise in AI applications has played a crucial role in the company's success.

The Challenges : Model Drift and Model Collapse

AI technology is facing significant challenges despite the industry's rapid growth. Leonard Lee and Karl Freund, analysts at neXt curve, discussed issues such as model drift, model collapse, and hallucinations during a webcast. These issues are critical obstacles to ensuring the reliability of AI applications.

The steep learning curve for AI is a major obstacle to adoption, particularly for generative AI, which is probabilistic in nature. A cautious approach is essential to integrating AI into mission-critical or safety-critical applications, which require high reliability. Karl Freund advised businesses to focus on AI’s practical capabilities by 2024. He stressed the importance of understanding what AI is capable of today while minimising associated risks.

Nvidia’s Response to AI Safety concerns

Nvidia has developed a new solution to address the challenges of AI, especially hallucinations. NeMo Guardrails is an open-source software tool that acts as a mediator in conversations with large language models (LLMs) to ensure they stay on track and relevant. This tool is a major step forward in AI safety. It offers a way to monitor and maintain the accuracy and context of AI-driven conversation.

NeMo Guardrails are customizable for a variety of applications including chatbots, finance, and healthcare. Nvidia contributes to the AI community by providing a way for developers implement safety measures.

Conclusion

AI's impressive market growth in 2023 and Nvidia’s important role in this expansion has been unmistakable. The challenges of model slippage, model collapse, and hallucinations highlight the need to continue efforts in order to improve AI reliability and safety. Nvidia's NeMo guardrails are a testament to industry commitment in addressing these issues. AI is evolving and it's important for companies to prepare for the future while harnessing current capabilities.

Image Source: Shutterstock

California, a state known for its technological innovations, has taken significant legislative steps to ensure ethical and safe deployments of Artificial Intelligence. The introduction by State Senator Steve Padilla of Senate Bills 892 & 893 is a significant step towards creating a robust framework to govern AI services, especially those that are contracted by state agencies.

Senate Bill 892 - A New Paradigm in AI Safety and Ethics

Senate Bill 892, a legislative proposal, mandates the Department of Technology develop comprehensive safety and privacy standards for AI services. The bill emphasizes that these standards must be mandatory for all AI companies contracting with state institutions by August 1, 2025. Sen. Padilla's move highlights the importance of a proactive approach to regulating AI. It acknowledges the significant impact AI has on the society, and the risks it could pose if not protected properly.

Senate bill 893: Fostering AI Research and Public Benefit

Senate Bill 893, in conjunction with Senate Bill 892, focuses on creating the California AI Research Hub. This initiative is a collaboration between the Government Operations Agency and the Governor's Office of Business and Economic Development. It also involves academic institutions. This bill's primary goal is to promote AI technology in California for the public good. It will ensure that AI research and developments are aligned with principles of privacy and security as well as societal benefit. The bill is designed to take advantage of California's position as a technology hub, and its world-class research universities, to democratize AI and promote innovation in the public interest.

California's Leadership in AI Regulation

California's AI legislative actions reflect a wider trend in the United States where states are increasingly recognising the need to regulate AI technology. Other states such as Texas, West Virginia and Louisiana have also taken steps to monitor or study AI systems that are used by state agencies. In the absence of federal regulation, this shows that state-level interventions are becoming more important.

Conclusion: Shaping AI's Future

Steve Padilla, a California state senator, introduced Senate Bills 892 & 893 to establish a framework for AI that will ensure that it is used ethically, safely, and in the best interest of society. California positions itself as an AI governance leader by setting standards for AI and creating a research environment which prioritizes the societal good. This proactive approach will ensure that AI technology development and usage aligns with ethical principles, and positively contributes to society.

Image source: Shutterstock