What’s behind AI 2.0?

Five enterprise-changing AI advances that you’ve (probably) never heard of

AI 2.0 is not one technology, but a collection of new technologies and techniques that are maturing simultaneously both in their technical feasibility and business applications. Here are five key technological advances that you probably haven’t heard about that are unlocking radical new AI capabilities.

1. Transformer Networks

What are they?
Giant pretrained, customizable, hyperaccurate, multitasking deep learning models


What are they used for?
Any hard problem with a significant time or context dimension (understanding and generating text, software code, etc.)


Who is using them?
Hyperscalers (Amazon Web Services, Google, IBM, and Microsoft), the advanced guard of speech and text analytics vendors, and many startups


2. Synthetic Data

What is it?
Generative models and simulated virtual environments used to create or augment existing training data


What is it used for?

Accelerating the development of new AI solutions, improving the accuracy and robustness of existing AI models, and protecting sensitive data


Who is using it?

Autonomous vehicles, financial services, insurance and pharmaceutical firms, and every computer vision vendor


3. Reinforcement Learning

What is it?
Machine learning approaches that test their way to optimal actions via simulated environments or a large number of microexperiments


What is it used for?

Constructing models that optimize many objectives/constraints or choosing actions based on positive and negative environmental feedback


Who is using it?
Firms targeting particular B2C marketing tasks, optimizing repeatable manufacturing processes, and robotic learning


4. Federated Learning

What is it?
A managed process for combining models trained separately on separate data sets


What is it used for?
Sharing intelligence between devices, systems, or firms to overcome privacy, bandwidth, or computational limits


Who is using it?
Hyperscalers, AI-enabled application vendors, and consumer electronics companies


5. Causal Inference

What is it?
Approaches such as structured equation modeling and causal Bayesian networks that help determine cause-and-effect relationships in data


What is it used for?
Business insights (e.g., attribution analysis) and bias prevention where insights and explainability are as important as prediction accuracy


Who is using it?
Innovation teams at leading organizations (e.g., determining treatment efficacy for a given disease at healthcare providers)

Read This Next

Avoid Zombie AI Projects With Our Best Practices

Download our AI myths and best practices guide. Learn how to avoid missteps, prioritize value, and move AI projects out of limbo with quick wins.