News

Joseph Caristena

Thu, 08 Oct 2026

Understanding the AI Landscape

What is AI ?

The term artificial intelligence usually refers to the ability of machines to mimic human cognitive functions without explicit programming. These functions encompass a wide range of capabilities, such as learning and problem-solving, but also include visual perception, speech recognition, and language translation. AI systems are used in many different contexts, from personal assistants like Siri and Generative AI tools like ChatGPT to advanced data analytics and self-driving cars.

Anchoring Creativity in Reality: The Importance of Grounding in GenAI

Do androids dream of electric sheep? The inclined science-fiction fan will recognize this question as the title of Philip K. Dick’s famous novel that served as the basis for the 1982 film Blade Runner. The central theme of the novel explores what it means to be human. Similarly, the existence of Large Language Models (LLMs) like ChatGPT or Google Gemini challenges us to consider what makes human-generated content unique and valuable, especially in the face of rapidly advancing GenAI tools.

There’s another interesting analogy between the novel and GenAI, specifically regarding theso-called "hallucination effect” of Large Language Models (LLMs).

This effect refers to the tendency of LLMs to generate text that appears plausible but may not be entirely accurate or factually correct. If asked for sources or citations regarding incorrect information, the LLM might even respond with fabricated citations or sources.

Similar to the replicants in the novel, who are indistinguishable from humans on the outside, LLMs can generate “realistic” text that seems to have been written by a human, but upon closer examination may lack the depth, creativity, and even fundamental reasoning powers of a human mind.

One of the key concepts used to minimize unwanted results like the hallucination effect is commonly known as grounding, which connects the creative output of GenAI models to the real world, enabling content output that is not only original but also accurate, relevant, and useful. For this to happen, GenAI has to not only access original training data to generate an output but it can also use external sources to make sure the desired factual context is present.