A new study by the Government Accountability Office (GAO) explores the opportunities and risks associated with generative AI. Examining the technology behind generative Artificial Intelligence and the many potential uses for these systems.
The report, entitled, Artificial Intelligence: Generative AI Technologies and Their Commercial Applications, is the first of many GAO reports delving into the world of generative AI.
Future reports will establish best practices, the societal and environmental effects of the using generative AI, and federal development and adoption of generative AI technologies.
This first installment provides an overview of how generative artificial intelligence works and how it differs from other kinds of AI, providing examples of its use across various industries, such as software engineering, medicine and education.
The report opens with a question – why does it matter? Going on to explain how enhanced capabilities and increased user interest has seen generative AI adopted by more than 100 million users.
A level of growth so significant it has generated much debate about its potential to revolutionize industries such as healthcare and education, versus the risks it presents to national security and the environment, and potential for spreading disinformation.
Key takeaways from GAO’s study:
- Generative AI differs from other AI systems in its ability to create novel content, the vast volumes of data it requires for training, and the size and complexity of its models.
- Generative AI systems employ several model architectures, or underlying structures. Referred to as neural networks, these systems are modeled loosely on the human brain and have the ability to recognize patterns in data.
- Commercial developers have created a variety of generative AI models that produce text, code, image, and video outputs, as well as products and services that enhance existing products or support customization and refinement of models to meet customer needs. Their benefits and risks are still unclear for many applications.
- A combination of factors enabled the rapid development of generative AI. Specifically, the availability of large datasets, refinement and augmentation of deep learning algorithms, and computer capacity.
- The training of generative AI models requires large amounts of data, commonly obtained from publicly available information on the internet, which can include copyrighted content.
- Commercial developers typically employ a process called reinforcement learning from human feedback, to provide more meaningful, fit-for-purpose responses. Humans evaluate and rank outputs, and then the models imitate the human preferences.
- Training large generative AI models can take tens of thousands of processors running for months and may cost several hundred million dollars.
