Accuracy
Overall correctness of predictions (i.e., how often the model is right).
Agentic AI
AI systems capable of autonomously planning, reasoning, and executing tasks across tools or systems to achieve goals. These systems can interpret instructions, generate plans, take actions, and monitor outcomes.
Algorithmic Bias
Systematic and repeatable errors in AI outputs that unfairly disadvantage certain groups.
API (Application Programming Interface)
A mechanism that allows different software systems to communicate with each other (e.g., accessing AI models via external services).
Artificial Intelligence (AI)
A broad field of computing focused on systems that perform tasks requiring human-like intelligence such as learning, reasoning, prediction, and decision-making.
Auditability
The ability to review, trace, and validate AI-generated decisions and processes.
Automation
The use of technology to perform tasks with minimal human intervention, typically focused on rule-based, repetitive work.
Backpropagation
Method used to train neural networks by working backward from the errors they make to adjust the model’s weights and improve its predictions.
Botscaling
Scaling an organization through AI systems, automation, and digital agents rather than increasing human headcount.
Closed Weights
AI models whose trained parameters are not publicly available; access is typically limited to APIs.
Cognitive Automation
Automation applied to knowledge work involving analysis, interpretation, and decision-making.
Data Governance
Policies and processes that determine how data is collected, stored, secured, and used.
Data Pipeline
The process of collecting, cleaning, and preparing data for use.
Data Privacy
The protection of personal and sensitive data, including how it is collected, stored, and shared in compliance with regulations and ethical standards.
Data Quality
The accuracy, completeness, consistency, and reliability of data used in AI systems.
Digital Labor
AI systems or software agents that perform work traditionally done by humans.
Digital Leverage
The amplification of organizational output through technology rather than workforce size.
Digital Transformation
The integration of digital technologies into business strategy and operations to improve performance.
Embedding
A way of turning text (or other data) into numbers so models can understand similarity.
Explainability
The ability to understand and interpret how an AI system produces its outputs.
F1 Score
A metric that balances precision and recall, useful when evaluating classification models with imbalanced data.
Fine-Tuning
Adapting a pre-trained AI model to improve performance on a specific task or dataset.
Foundation Model
A large AI model trained on broad datasets that can be adapted for multiple applications.
Frontier Firms
Companies that use advanced technologies (like AI) better than others to achieve superior performance.
Frontier Models
AI models with the most advanced, large-scale AI systems that feature state-of-the art capabilities. These models typically expand the functionality of existing large language models (LLMs) through enhanced reasoning, coding, and general intelligence. They are often used to accelerate breakthroughs in science, biotech, etc. These large models require large amounts of data, computation power, and human expertise.
Generative AI
AI systems that create new content (e.g., text, images, code) based on learned patterns.
Generalization
The ability of a model to perform well on new, unseen data.
Ground Truth
The actual, correct outcomes used to train and evaluate models.
Hallucination
When AI generates incorrect or fabricated information presented as factual.
Human Advantage
Uniquely human capabilities such as judgment, creativity, empathy, and ethical reasoning.
Human–AI Collaboration
The interaction between humans and AI systems where each contributes complementary strengths.
Human–AI Enterprise
An organization where humans and AI systems work together to produce outcomes neither could achieve alone.
Human-in-the-Loop
A design approach where humans review or validate AI outputs, especially in high-stakes decisions.
Inference
Using a trained model to make predictions on new data.
Label (Target Variable)
The outcome the model is trying to predict (e.g., turnover, promotion).
Large Language Model (LLM)
A type of generative AI trained on large volumes of text data to understand and generate natural language.
Machine Learning
Algorithms that learn patterns from data to make predictions or decisions.
Model
A trained algorithm used to make predictions or decisions.
Model Deployment
Putting a trained model into real-world use (e.g., inside an HR system).
Model Drift
The degradation of model performance over time due to changing real-world conditions.
Model Training
The process of teaching a model to recognize patterns using data.
Natural Language Processing (NLP)
AI techniques used to analyze and interpret human language (e.g., text or speech).
Neural Networks
Computer systems designed to learn patterns from data and make predictions or decisions, inspired by how the human brain works.
Open Source AI
AI systems where architecture and training code are publicly available for use and modification.
Open Weights
AI models where trained parameters are publicly available, but training data or methods may not be fully disclosed.
Overfitting
When a model performs well on training data but poorly on new data.
Pivotal Talent
Roles that create disproportionate strategic value, often including AI, data, and translation roles.
Precision
Of all predicted positive outcomes, how many are actually correct.
Predictive Analytics
Using data and models to forecast future outcomes.
Prescriptive Analytics
Recommending actions based on predictions and optimization models.
Prompt
The input or instruction given to a generative AI system.
Prompt Engineering
Designing prompts to improve the quality and relevance of AI outputs.
Recall
Of all actual positive outcomes, how many were correctly identified by the model.
Reinforcement Learning
A type of machine learning where models learn by receiving feedback (rewards or penalties) based on actions.
Responsible AI
The practice of designing and deploying AI systems that are fair, transparent, accountable, and ethical.
Responsible Talent Algorithms
AI systems used in HR decisions that are designed to ensure fairness, compliance, and defensibility.
Revenue per Employee
A productivity metric calculated as total revenue divided by number of employees; often higher in AI-enabled firms.
Selection Algorithms
AI-based systems used in hiring and talent selection.
Sentiment Analysis
Analyzing text to determine emotional tone (e.g., employee feedback).
Signal vs. Noise
The distinction between meaningful information and random variation in data.
Skills-Based Organization
An organization that deploys talent based on skills and capabilities rather than job titles.
Solo Unicorn
The idea that AI could enable a single individual to build a billion-dollar company using automation and digital infrastructure.
Supervised Learning
A machine learning approach that uses labeled data to predict known outcomes.
Task-Level Automation
Automating specific tasks within a job rather than replacing the entire role.
Test Data
New data used to evaluate how well the model performs.
Training Data
The data used to teach a model.
Transparency
Clarity about how AI systems are used and how decisions are made.
Underfitting
When a model is too simple and fails to capture important patterns.
Unsupervised Learning
A machine learning approach that identifies patterns in unlabeled data.
Use Case
A specific application of AI to solve a business problem.
Work Decomposition
Breaking jobs into tasks to determine what can be automated, augmented, or performed by humans.
Work Redesign for AI
Reconfiguring tasks, roles, and workflows to maximize AI’s benefits.