The Predictability Pandemic

Context

• As Large Language Models (LLMs) and predictive AI keyboards (such as Gboard, Microsoft SwiftKey, and generative chatbots) become integrated into daily human communication, data scientists and cognitive psychologists have raised critical flags regarding “The Predictability Pandemic.”
• The widespread adoption of predictive algorithms is shifts human interaction from an active, creative construction of language to a passive selection of machine-generated suggestions. This shift introduces significant risks, notably Model Collapse, a systemic threat to the future training architecture of advanced artificial intelligence models.

1. What are Predictive Keyboards?

Predictive keyboards are AI-based software systems that:

• Predict the next word or phrase while typing.
• Learn from user behaviour and communication patterns.
• Use machine learning and natural language processing (NLP).

Examples:

• Smartphone autocorrect systems
• AI writing assistants
• Email smart replies
• Chat-based AI systems

2. Major Side Effects

I. “Model Collapse” (The AI Threat)

• What it means: AI models learn how to write by reading human text on the internet.
• The Problem: As more people use AI or smart keyboards to write, the internet gets filled with generic, AI-generated text. When future AI models learn from this artificial text instead of real human writing, they start repeating the same mistakes and eventually break down. This is called Model Collapse.

II. Cognitive Offloading & “The Google Effect”

• What it means: Our brains like to save energy. When we know a tool (like a calculator, GPS, or predictive text) can do the job for us, our brain stops trying to remember it.
• The Result: If you know Google or your keyboard will finish a fact or sentence for you, your brain stops practicing how to remember or build those thoughts from scratch.

 III. Losing Our Unique Vocabulary

• What it means: Humans have “emotional granularity,” which means we can pick precise words to describe exactly how we feel (like choosing between gloomy, bitter, or exhausted).
• The Result: Smart keyboards prefer common words. They constantly suggest generic phrases like “I’m fine” or “I’m sad”. If we keep clicking them just to save time, our active vocabulary shrinks, making our language boring and flat.

With reference to “The Predictability Pandemic” and predictive AI systems, consider the following statements:
1. Predictive keyboards use Machine Learning and Natural Language Processing (NLP). 
2. “Model Collapse” refers to the breakdown of AI quality when models increasingly learn from AI-generated content instead of original human-created data. 
3. The “Google Effect” suggests that frequent dependence on digital tools can reduce human memory retention and cognitive effort. 
Which of the statements given above are correct?
(a) 1 and 2 only
(b) 2 and 3 only
(c) 1 and 3 only
(d) 1, 2 and 3
Answer:
(d) 1, 2 and 3
Explanation:
•  Statement 1 is Correct: Predictive keyboards function using Machine Learning (ML) and Natural Language Processing (NLP) to predict words, phrases, and communication patterns. 
•  Statement 2 is Correct: Model Collapse occurs when AI systems increasingly train on AI-generated content instead of authentic human-created data, leading to repetitive outputs, reduced originality, and degradation in model quality. 
•  Statement 3 is Correct: The Google Effect refers to cognitive offloading, where humans rely on digital tools (like Google, GPS, calculators, or predictive keyboards) instead of actively remembering information or constructing thoughts independently.