Generative artificial intelligence

Sci-Tech Neologisms-4.

Photo by Ricardo Gomez Angel on Unsplash

Parag Mahajani

What is generative artificial intelligence?

• Sub-discipline of AI. Designed to mimic the human brain.
• Applies ML algorithms, neural networks, and DL techniques.
• Powered by generative adversarial networks (GANs) and large language models (LLMs) called as foundation models.
• Trained on large dataset samples, GenAI recognizes the deeper patterns and relationships within the data to create mathematical structures known as latent space.
• GenAI generates new content based on latent space and statistics.
• GenAI refines its output based on more and more data over time.

Why is it used?

• Produces synthetic content: text, images, video, audio, and code.
• Employ transformer-based large language models (LLMs) for NLP-based applications, content generation, and summarization.
• Based on multimodal AI models.
• Uses generative adversarial networks (GANs) for realistic images, video, and audio content.
• Employ variational auto-encoders (VAEs) to create original-looking images, video, and audio content.
• Optimizes costs due to increased automation of business processes.
• Faster decision-making using advanced analytics and predictive intelligence.
• Accelerate time to market (services and products).

What are its current challenges?

• Fear of discrimination: Businesses should follow best practices that promote fairness, transparency, and social responsibility.
• Ethical concerns: Businesses should ensure full accountability and transparency to avoid ethical potholes.
• Tech complexity: Businesses need large computing resources and advanced data processing capabilities. Power requirements can skyrocket.
• Hallucination: Businesses should provide more accurate training data to avoid inaccurate and fictional responses from GenAI tools.

Deep dive

Generative AI Defined: How it Works, Benefits and Dangers

The Indus project

Disclaimer: The statements and opinions expressed in this blog are those of the author(s) and do not necessarily reflect the positions of Thoughtworks.