GoodAI Research

“GoodAI is building towards my lifelong dream to create general artificial intelligence. I’ve been focused on this goal since I was 15 years old.”

Marek Rosa, CEO, CTO, & Founder

To date, most of what we consider general AI research is done in academia and inside big corporations. GoodAI strives to combine the best of both cultures – academic rigor and fast-paced innovation – to create the right conditions for cooperation across boundaries.

Current Projects

Long-Term Memory

We are developing Long-Term Memory (LTM) systems to expand LLM agents' context window, enabling continual learning from interactions and environmental dynamics. Our goal is to create agents capable of lifelong learning, using every new experience to enhance future performance.

Charlie Mnemonic

We are open-sourcing Charlie Mnemonic, the first personal assistant (LLM agent) equipped with Long-Term Memory (LTM). The distinctive feature is the implementation of LTM, enabling it to learn from every interaction.

LLM Agents

At GoodAI, we are dedicated to pushing the boundaries of artificial intelligence. Our current focus is on the development of Large Language Model (LLM)-based agents with personalities that go beyond simple conversations, and instead exhibit LLM-driven behaviors, interacting with humans and other agents, as well as their virtual environment.

GoodAI Groundstation

At GoodAI, we are building collaborative AI agents that enhance human capabilities and drive positive change at scale. Collective intelligence is the guiding principle behind our work.

Ethics & Policy

Mitigating risk. Our work on AI’s long-term impacts on society. With narrow AI becoming ever more powerful, and the possibility of general AI on the horizon, it is vital that adequate impact assessment and governance models are put in place...

GoodAI Grants

GoodAI has a history of awarding grants to research projects that advance the field of artificial intelligence.

Past Initiatives

Badger Architecture

Badger Architecture is the unifying framework for our research. The main principle of Badger is modular life-long learning. The modular principle is expressed in the architecture: a network of identical agents - Badger experts.

GoodAI Roadmap

In 2021/22 GoodAI will focus on four core research areas: learning to learn, lifelong (gradual) learning, open-endedness, and generalization / extrapolation of meta-learned algorithms.

Badger Seminars

We run the series of Badger Seminars to bring together world-renowned experts in order to advance the field of life-long learning, meta-learning, multi-agent learning and other directions that we believe are relevant for building generally intelligent agents.

Publications

We view intelligence as a tool for searching for solutions to problems. The guiding principles of our AI research revolve around an AI which can accumulate skills gradually and in a self-improving manner. Each new skill works like a heuristic...

Collaborations

ICLR 2022: From Cells to Societies

GoodAI researchers co-organized a workshop with the University of Chemistry & Technology in Prague (VSCHT) and Google Research for the 2022 ICLR Conference. The workshop examined both natural and artificial systems to see how they can (or already do) lead...

VeriDream

Advances in artificial intelligence (AI) are changing the business models of many companies, and creating entirely new ones. The VeriDream project, an international consortium of six organizations across Europe, was awarded €2 million by the European Innovation Council to carry...

IV4XR

“Extended Reality” (XR) systems are advanced interactive systems such as Virtual Reality (VR) and Augmented Reality (AR) systems. They have emerged in various domains, ranging from entertainment, cultural heritage, to combat training and mission critical applications. The development and authoring...

AI Challenge

Global citizen science project tackling crucial research problems in human-level AI development.

HLAI Conference

In August 2018 the Human-Level AI Conference came to Prague. The conference was created by Tarek R. Besold as a combination of three established conferences AGI, BICA and NeSy.