Issue link: https://digital.shearman.com/i/1512772
Shearman & Sterling LLP Leveraging the Broad Potential of Artificial Intelligence While Mitigating the Risks | 6 Generative AI Generative AI refers to a class of artificial intelligence that has the ability to generate new content, such as text, images, audio, or other data, which is comparable to what a human might generate. Utilizing deep learning networks, Generative AI systems are able to dissect data and understand patterns and structures to produce new and novel content that is built upon and similar to existing data on which it is trained. Large Language Model Large Language Model (LLM) is an example of Generative AI—indeed, the type of Generative AI creating the most buzz lately—that produces human- like text and natural language. LLMs are trained to detect patterns and connections using large datasets of texts (including, e.g., books, articles, or other materials available on the internet, such as social media posts). These models have the ability to predict the next word in a sentence and generate a coherent text by understanding and replicating patterns, context, and grammar from the training data. For example, models that are trained on the works of William Shakespeare can easily generate a college admission essay in the style of Shakespearean English. HOW IS AI USED? For years, companies have benefited from using artificial intelligence tools to improve efficiency, save time, and decrease costs. Some examples of existing AI applications include customer service chatbots, product recommendation systems (common in e-commerce, social networking, and streaming services), and fraud detection in finance and cybersecurity. With Generative AI, companies stand poised to reap even greater benefits as the technology develops. Generative AI tools can save time and resources by instantly producing written content, such as articles, product descriptions, simple or form contracts, and marketing materials. These tools can also be used to supplement a company's decision-making process by analyzing extensive amounts of data to identify patterns and trends, make predictions, and generate alternative scenarios. UNDERSTANDING AND MITIGATING THE RISKS OF AI While AI offers many applications and services that can improve operations and increase efficiency, assessing and prudently implementing these technologies requires careful and comprehensive consideration of both macro risks (like the rapidly evolving regulatory environment) and company- specific risks. To capitalize on AI while managing these risks, we recommend that companies consider forming oversight bodies tasked with (a) conducting diligence on AI, (b) advising management regarding the adoption of AI technology, and (c) developing policies, procedures, and other safeguards to ensure that risks are assessed and appropriately mitigated. The suggestions discussed herein are general guidelines; while these recommendations are a starting point for consideration, every company is different, and a one-size-fits-all approach can overlook key risks unique to your company. We therefore suggest that you customize your strategy to align with the needs of your business. Establishing an AI Task Force Companies should consider forming an AI governance committee or task force to ensure that AI technologies are properly vetted and deployed, effectively, in line with the company's broader strategic vision and ethically. • Composition of the task force. As discussed below, the AI task force should be responsible for evaluating the existing technologies, assessing how they fit within the company's business model, considering the legal and regulatory requirements (or restrictions) on the technologies, and assessing the risks. Therefore, an effective AI task force must be cross-disciplinary, including business leaders, compliance, legal, and IT. The company should consider the appropriate reporting structure for the task force, and how the work performed by the task force can be used to further management and board- oversight of the company's use of AI.